a nationwide qi project towards mdg 4

7/31/2019 A Nationwide Qi Project Towards Mdg 4

1/30

Title: A nationwide quality improvement project to accelerate Ghanas

progress towards Millennium Development Goal Four: design and

implementation of innovation and scale-up

Authors: Nana A. Y. Twum-Danso,1 George B. Akanlu,2 Enoch Osafo,2 Sodzi Sodzi-

Tettey,1 Richard O. Boadu,2 George A. Adjei,2 J. Koku Awoonor-Williams,3 Alexis

Nang-Beifubah,3 Akwasi Twumasi,3,3 C. Joseph McCannon,1 Pierre M. Barker1,4

1 Institute for Healthcare Improvement

20 University Road, 7th Floor

Cambridge, MA 02138

USA

2 National Catholic Health Service

P. O. Box KA 9712

Airport, Accra


2/30

University of North Carolina at Chapel Hill

Chapel Hill, NC 27516

USA

Corresponding Author:

Pierre M. Barker MD

Institute for Healthcare Improvement

20 University Road, 7th floor

Cambridge, MA 02138, USA


3/30

The gap between evidence-based guidelines and practice of care is particularly

evident in low- and middle-income countries, as reflected by high maternal and child

mortality rates. We designed a phased, rapid, national scale-up quality improvement

(QI) intervention to accelerate the achievement of Millennium Development Goal Four

in Ghana.

Methods

We used QI approaches that emphasize systems thinking, motivation of frontline

providers, generating and testing of change ideas and learning from data at the local

level, transparent data reporting, and sustainability. Working within an adaptable

framework to address the underlying drivers of child survival, we redesigned

implementation strategies in an iterative manner based on the lessons learned.

Results

After 50 months of implementation, we have completed two prototype learning

phases, each of 18-month duration, and have begun regional spread phases that

cover all 38 districts of the three northernmost regions of Ghana, serving a population


4/30

prototype phases for innovation and learning by frontline health staff, and scaling up

rapidly over time and space with support from their managers and senior leaders.

Keywords: quality improvement, health systems strengthening, large-scale

improvement, maternal newborn and child health, Millennium Development Goal

Four, low-resource setting


5/30

Introduction

The gap between evidence-based guidelines and practice in health care is

particularly wide in low- and middle-income countries (LMICs), where maternal and

child mortality rates are high despite the availability of cost-effective interventions .

In 2010, only 19 of 68 countries were on track to achieve Millennium Development

Goal Four (two-thirds reduction in mortality in children less than five years old (Under-

5) from rates in 1990 by 2015) . Efforts to implement child survival programs in sub-

Saharan countries have not had the expected effect of reducing child mortality (7).

Even when pilot projects are well implemented, they are rarely designed with scale-

up as an explicit design strategy. Thus, many may scale up well in environments that

are similar to the test circumstances, but are challenged by limited ability to adapt to

local context. Likewise, unless designed carefully, scale-up may be challenged by the

need for additional resources and the lack of a mechanism to sustain the intervention.

Other scale-up design considerations include supervision, monitoring, evaluation, and


6/30


7/30

Design

Quality Improvement Approach

The QI methods used in this project are based on well-described health system

improvement approaches . Starting with a hypothesis that we could improve the

drivers of preventable causes of death for children Under-5 (Figure 1), we

challenged the status quo through development of transparent data reporting

systems, generation and testing of local ideas for improvement, scale-up designs

that rely on rapidly spreading locally-tested successful innovations, use of existing

resources, and developing local capacity. This improvement approach emphasizes

systems thinking, motivation of frontline providers and their managers,

contextualization of implementation strategies, analysis and learning from data at

the local level, redesigning strategies in an iterative manner based on lessons

learned, reliability principles, local ownership, and sustainability .


8/30

members of the ICN. The QI teams were multidisciplinary and varied in number from

four to 10. In health centres, the QI team was typically led by a midwife, while in

hospitals, it was typically a doctor.

At Learning Sessions, project facilitators taught QI methods and helped health staff

analyze their local health systems and processes, identify process failures or

implementation gaps and reasons for them, and develop and plan the testing of

specific changes that they believed were likely to lead to improvement, using the

Model for Improvement as their guiding framework. In between Learning Sessions,

project facilitators accompanied Change Agents - typically public health nurses,

disease control officers, and health information officers from the regional, district,

and diocesan health management teams - to regularly visit each QI team in their

facility. The facilitators and change agents coached teams to test their changes,

analyse their data to assess whether their changes were resulting in improvement,

and developing new changes to accelerate and sustain improvement. The successful


9/30

integrate these visits with other district field supervisory activities. Additional

technical support was provided by telephone between site visits.

An explicit data quality improvement (DQI) component was also designed into the

project to improve the routine health information system (RHIS) of the GHS. The

health information officers in the hospitals and the district health offices formed a

separate ICN focused on DQI. They were trained and coached in general QI methods

as well as specific DQI strategies focused on the completeness, timeliness, and

accuracy of the data collected and reported in the RHIS.

Capacity Building in Quality Improvement

Capacity building in QI methods and facilitation skills was provided in a phased

manner. In Phase 1, the project staff received an introduction to QI. This was followed

by a 10-month in-depth longitudinal professional development course in QI,


10/30

1. Increase in the number of project staff supporting the QI teams in Phase 2 by

only three-fold while scaling up the number of QI teams by more than 10-fold

as we relied more heavily on the Change Agents to implement and sustain the

QI intervention;

2. Joint facilitation of Learning Sessions and site visits between the project staff

and the Change Agents;

3. Convening of Learning Sessions by the regional, district, and diocesan health

leaders independently of the project staff;

4. Quarterly review meetings for the Change Agents, led by the regional health

leadership team, to review progress, successes, and challenges;

5. Funding and integration of Change Agents QI coaching site visits into routine

monitoring and supervision of their MNCH work;

6. Inclusion of QI presentations and discussions in the agenda of the already

established district and regional health twice-yearly performance review


11/30

Phase 0 An eight-month assessment and planning phase, during which we

collected detailed baseline MNCH performance data and identified early

adopter sites for the prototype phase (Phase 1). A crucial period of will-

building, where the project was explained in detail to regional, district, and

diocesan health leaders, ensued.

Phase 1 Prototype phase including 25 health centres and two hospitals across

three rural districts and one Catholic diocese in the three northernmost regions

of Ghana, to test and develop a change package to improve MNCH care

processes for use in the scale-up phases.

Phase 2 Initial scale-up phase including all the health facilities (both public

and private sector) in the 38 districts of the north (population: five million), in

addition to the three districts in Phase 1.

Phase 3 To develop a change package to improve care for infants and

children admitted to hospital, and test the unique role of a faith-based health

system in a nationwide improvement initiative we enrolled all the Catholic


12/30

Using Everett Rogerss model of the diffusion of innovations , we theorized that the

project staff would be the innovators, while the early adopters would be those

health staff in the prototyping phases (i.e., Phases 1 and 3a). The rest of the health

staff/facilities would represent the early majority and the late majority (i.e.,

Phases 2, 3b, 3c and 4). The small-scale prototype phases were designed to achieve

the objectives of demonstration, advocacy, and building more will for the next phase.

Evaluation

Independent evaluation of this project was undertaken by a third party, which worked

in a facilitative but independent manner from the project implementation team. Their

evaluation, which covers the changes in process performance and outcomes will be

reported elsewhere.

Data Collection


13/30

2. Detailed description of the specific change ideas tested by each QI

team, including the dates testing began and ended, dates on which change

ideas were modified, and observations and insights gained from the testing.

These were documented by the project staff in notebooks during site visits and

transferred subsequently to an electronic database, termed a change

tracker.

3. Health process indicators reflecting the changes in the care processes for

pregnant women and the Under-5 in both outpatient and inpatient settings.

The majority of these were already being collected and reported to the RHIS,

although a few new indicators (e.g., interval between identification of a sick

child and initiation of definitive treatment) were developed. Most of the new

indicators were in the clinical registers, while a few had to be developed de

novo. QI teams also collected context-specific indicators based on the change


14/30

4. Health outcome indicators focused primarily on mortality, not morbidity,

because the former were more likely to be present in the RHIS. Based on

specific change ideas being tested, several QI teams collected morbidity data

or case fatality data that were not required for the RHIS but were needed to

determine whether change ideas were leading to improvement. In addition, we

collected data from GHSs community-based surveillance volunteer data which

captures births, deaths, and notifiable diseases that occur at the community

level; these data are reported monthly by the volunteers to the health centres.

All the data reported into the RHIS are summarized at the sub-district, district, or

regional level and are de-identified as are the other data collected by the QI teams,

community volunteers, and project staff.

Data Analysis


15/30

tracker were used to confirm the exact nature of the change, when it was initiated

or modified, and when it was terminated.

Ethics

No institutional review board approval was required for this work, as the

implementation and monitoring of the QI interventions to improve the MNCH

program were considered part of the established and ongoing MNCH program of the

GHS which is based on existing policies of the Ghanaian Ministry of Health. Program

evaluation used routinely collected de-identified aggregate data of process

performance and outcomes.

Results


16/30

The first eight months of the project included a major focus on building will. We

established the partnership between IHI and NCHS, and strengthened the existing

partnership between NCHS and GHS at the corporate level. We introduced the project

design to stakeholders at all levels of the health system. We visited hospitals and

clinics and spoke to the health staff to ensure that we understood contextual factors

affecting the quality of health care, supervision and management at the facility,

district, and regional levels, and how the QI intervention could improve upon it.

After launching the project, we continued to build and maintain will through regular

feedback sessions with the frontline providers and their managers after each round of

coaching visits and Learning Sessions. We disseminated the results of the project at

GHSs performance review meetings at district, regional, and national levels, as well

as at national and international conferences. Four months before launching Phase 2,

the project held a dissemination meeting for the three northernmost regions, sharing

results to date and discussing readiness for scale-up. In Phase 3, the NCHSs annual


17/30

and then tested and analyzed during the innovation phase of the project. By the end

of Phase 1, the 27 QI teams had tested 104 specific change ideas, while by the end of

Phase 3a, the nine hospital QI teams had tested 47 specific change ideas. These

change ideas spanned pregnancy identification and registration for antenatal care

(ANC) in the first trimester of pregnancy, receiving quality care during at least for four

ANC visits, accessing skilled delivery on time and receiving quality perinatal care and

postnatal care, early care-seeking for sick children and women in labor, triage,

adherence to protocols for the most common causes of childhood illnesses in their

local context, and DQI protocols.

We developed the description of drivers that we hypothesized would lead to better

outcomes in the Under-5 population. Throughout the project we sought to better

understand the contextual factors underlying preventable deaths in the Under-5s in

both the community and health facilities. We summarized these factors into a

construct known as a driver diagram, which we then populated with broad change


18/30


19/30

subsequent Learning Sessions. To date, we have provided eight separate QI training

and coaching sessions for district Change Agents. The DQI initiative with the health

information officers in Phase 2 began five months after the launch of Phase 2. As of

December 2011, 26 out of 38 (68%) districts were actively working on improving the

accuracy, completeness, and timeliness of the data reported to them by the frontline

providers in the clinics and hospitals.

Phase 3a was launched in the nine worst-performing hospitals in the NCHS (see Table

1) in October 2009 and ended in April 2011 (18 months duration), with the

development of a second change package (Table 3) using the same approach as in

Phase 1. We promoted this change package to the remaining 20 NCHS hospitals in

the south in Learning Sessions grouped according to three Catholic provinces over the

course of six weeks starting in June 2011. We also promoted this change package

to all 36 hospitals in Phase 2 during Learning Sessions which began in June 2011. As

of December 2011, all 29 hospitals in this scaled-up phase of the NCHS, Phase 3b,


20/30

Our approach, which emphasized creating of partnerships across the health sector,

engaging frontline providers and their managers and senior leaders, and

demonstrating and disseminating early results regularly to all stakeholders, served to

build will, secure buy-in for the current work, enhance preparation for subsequent

phases of the project, and engage in policy dialogues at all levels of the health

system. Our intervention drew heavily on change ideas generated by frontline

providers, taking account of local context and working within the boundaries of the

national MNCH program. Frontline staff and managers were empowered to test those

changes and learn from them, resulting in the assembly of locally proven adaptable

change packages that could be credibly spread to exponentially larger numbers of

frontline providers working in similar and different contexts during the scale-up

phases, using Everett Rogerss model for diffusion of innovations .

In addition to the project execution design, two QI methods evolution of adaptive

driver diagrams and development of change packages proved to be powerful tools


21/30

The projects explicit aim for large-scale implementation was conceptualized right

from inception. We embedded a sustainability strategy into the design that included

ownership of the change process by regional, district, and diocesan leaders,

development and coaching of large numbers of local health system supervisors

(Change Agents). Through a rapid weaning strategy, we progressively decreased

reliance of Change Agents on project staff (e.g., increasing the hand-over of

independent site visits as the Change Agents became more skilled in QI and

facilitation).

Iterative learning and testing of the project design itself was a key part of executing

this project. In addition to the continuous internal assessment of results, which has

allowed for design flexibility and rapid improvement of performance, a formal

evaluation of the project is underway, undertaken by an external evaluator, using

quasi-experimental techniques .


22/30

paper, as the project design continues to iterate and the full benefit of this design is

yet to be realized. Another limitation is the focus on the progress of implementation

of the project design without inclusion of health process and outcome data; as such,

we have yet to demonstrate that this QI intervention is on track to achieve its

intended aim of accelerating the achievement of MDG 4 in Ghana. A formal evaluation

is underway, and interim process and outcome results will be reported shortly.

Acknowledgements

We would like to express our gratitude to Patrick Ansu and Linda Azumah for data

collection and analysis and to Jane Roessner for copy-editing.


23/30


24/30

References


25/30

Figure 1. Conceptual framework of the underlying drivers of preventable deaths in children Under-5 in Ghana

Legend: NHI=National Health Insurance

25


26/30


27/30

Table 1. Implementation progress of Phases 1, 2, 3a and 3b of the QI interventions

Phase(timing)

Health Facilities No. ofdistric

ts

Learning Sessions asof December 2011

Joint Site Visits as ofDecember 2011

No. ofhealth

posts/centres

No. ofhospitals Total

Total no.of

sessions

Mean no.

ofparticipants persession

Total no.of visits

Mean no.of visitsper QIteam

Phase 1(07/2008 to

12/2009) 25 2 27 3 10 85 394 14.6Phase 2

(09/2009 topresent) 540 36 576 38 35 54 1424 5.5

Phase 3a(10/2009 to04/2011) 0 9 9 4 35 77 8.6Phase 3b

(06/2011 topresent) 0 29 29 3 47 54 1.9

27


28/30

Table 2. Summary of changes found to be effective in improving processes of care

for the antenatal, perinatal and postnatal periods after Phase 1

Care Pathway Successful Change Idea(s) Facility

Type*

ANTENATAL1. Regi

stration in

1st

Trimester

1A. Community stakeholder meetings with opinion leaders and other

influential groups about the importance of early and regular ANC

1B. Community stakeholder meetings followed by registration of

pregnant women by community volunteers on monthly basis

C

C

2. At

least 4

visits

before

delivery

2A. Increase number of days ANC is offered at static site AND re-

design clinic processes to reduce visit duration per client to < 1hr

2B. Offer ANC as outreach service as well as at static site AND re-

design clinic processes to reduce visit duration per client to < 1hr

C & H

C

PERINATAL3. Skill

ed

Delivery

&

Immediat

e

Postnatal

Care

3A. Video show in communities on the risks of labour & delivery

3B. Male advocacy group in communities to promote skilled delivery

3C. TBA engagement on risks of unskilled delivery and provide

incentives3D. Use ANC register to identify women at 36+ weeks gestation for

home visits to remind them & family members about skilled delivery &

confirm transport plan

3E. Provide domiciliary delivery if, upon notification by mobile phone,

labour too advanced, woman has no means of transport from

community or health staff cannot arrange transport from clinic or

hospital

3F. Create a welcoming, patient-friendly environment in health facility

for labouring women

3G. Create systems to ensure consistent and correct use of

C

C

C

C

C

C & H

C & H

C & H


29/30

5C. If woman lives in distant community without CHO AND return

facility visit not possible AND health staff home visit not possible, train

IMCI volunteers to provide Day 6/7 care.

Legend: *C=Health centre, clinic or health post; H=hospital


30/30

Table 3. Summary of changes found to be effective in improving processes of care in the hospital outpatient and

inpatient setting after Phase 3a

Driver Area of

Clinical/

CommunityCare Change Concept

Package # Description of Successful Change Ideas

Delay in

Seeking

Care

Care-

seeking

behaviour

Targeted health

education

1A Targeted health education on early care-seeking using interactive

platforms (e.g. radio)

1B Community engagement and education via durbar or place of worship

ReferralEngaging primary

providers1C Engagement with health providers (both traditional and allopathic) on need

for early referral and early warning signs

Prompt

Diagnosis

and

Treatment

Triage

2A

Triage system for screening and emergency treatment of critically ill

children

Separate Under-5 OPD services from adult OPD service

Prioritize Under-5 outpatient care

Prioritize Under-5 inpatient care

Delay in

Providing

CareFast Track

Non-

Adherence

to

Protocols

Adherenceto Protocols

Training/Coaching/

Mentoring

3A Training staff on protocols followed by regular coaching and mentoring

which include ad hoc testing on site with immediate feedback

3B Training postpartum women and other care givers on hygienic cord care

through demonstration, practice and immediate feedback

3C Mother-to-mother support group on food choices and frequency of feeding

while on admission under mentoring of nurses

Task-shifting 3D Empowering nurses to start acting on standard treatment protocols before

doctor arrives

30

a nationwide qi project towards mdg 4

Documents