Effect of audit and feedback on the availability, utilisation and quality of emergency obstetric care in three districts in Malawi

Audit and feedback 

Audit and feedback is a broad term which has been defined as “any summary of clinical performance of health care over a specified period of time”.7 In maternity care audit and feedback may refer to either maternal death audit, perinatal death audit, near-miss case review or criterion-based audit. The term maternal death audit is a broad term that describes different approaches used to study maternal deaths with view to reducing future maternal mortality and morbidity by improving the quality of care. These approaches are confidential enquiries into maternal deaths, maternal death surveillance, facility-based death reviews and community-based death reviews (also called verbal autopsy).8

Criterion-based clinical audit on the other hand is a more specific term which has been defined as “a quality improvement process that seeks to improve patient care and outcomes through systematic review of care against explicit criteria and implementation of change. Aspects of structure, processes, and outcomes of care are selected and systematically evaluated against explicit criteria. Where indicated, changes are implemented at individual, team, or service level and further monitoring is used to confirm improvement in healthcare delivery”.9 Criteria-based audit consists of five steps that form the classic audit cycle: establish standards of good practice, measure current practice, feedback findings and set local targets, implement changes in practice where indicated, and re-evaluate practice and feedback.8

Both maternal death review and criterion-based clinical audit are supported by expert opinion and have recently been endorsed by the World Health Organisation.8 A Cochrane systematic review that included 118 randomized controlled trials concluded that criterion-based audits can bring about improvements in professional practice.7 A second Cochrane review set out to evaluate the effect of maternal and perinatal death reviews on maternal and perinatal mortality, but did not find any randomized controlled trials.10 However, observational studies have shown that maternal death reviews can improve practice and reduce maternal mortality.11

Design 

We conducted an observational study in which the availability, utilisation and quality of emergency obstetric care at baseline (2005) was compared with Year 1 (2006) and Year 2 (2007). Maternal death review and criterion-based audit were introduced in the three districts in 2006.

Setting 

The study was conducted in three districts (Lilongwe, Kasungu, Salima) with a total population of 2,944,360. There were a total of 92 health facilities in the three districts – this includes 13 hospitals, 60 health centres and 19 dispensaries. In general dispensaries do not provide maternity services apart from family planning (FP) services. All 73 health facilities (13 hospitals and 60 health centres) that provide maternity care were included in this study; 18 in Kasungu district; 42 in Lilongwe district and 12 in Salima district. Permission for the study was obtained from the Malawi Ministry of Health. Only 1 out of 60 health centres provides all the six signal functions of a fully functioning BEmOC facility; the rest of the health centres provide only some of the six signal functions.12

Intervention 

The programme was launched in February 2006, and maternal death case review and criterion-based audit were introduced in the three districts in August 2006.[13], [14]

The Malawi Ministry of Health developed three forms which are currently used for maternal death review: (a) maternal death notification form contains the particulars of the deceased and its purpose is to notify the District Health Office within 7 days of the maternal death, (b) maternal death review form is completed during maternal death review meetings and contains details of the causes of maternal death, factors that contributed to the maternal death, and recommendations made during maternal death review, and (c) maternal death follow-up form is used to follow-up the implementation of recommendations made during the maternal death reviews. Using interactive quality improvement workshops, health care providers working in maternity units and hospital administrators in the three districts were trained on how to conduct maternal death reviews in a “blame-free” manner. Standardization in training was ensured by conducting joint workshops that brought together participants from the three districts. Each hospital set up a quality improvement team (made up of health professionals working in the maternity unit [midwives, clinical officers, doctors], laboratory technicians, anaesthetic technicians and administrators). The team met whenever there was a maternal death and reviewed the case to identify the cause of death and factors that contributed to the death, and made recommendations for change. At the beginning of each maternal death review session, the team reviewed whether the recommendations made during previous sessions were implemented and (where applicable) why they were not implemented.

The criterion-based audit intervention consisted of the classic five-step audit cycle: (a) established standards of good practice, (b) measure current practice, (c) identify gaps in current practice and feedback to health professional, (d) implement recommendations for change, and (e) re-evaluate practice and recommendation. Using interactive workshops the health care providers and hospital administrators were trained on how to conduct criterion-based audit. Standardization in training was ensured by organizing joint workshops attended by participants from the three districts. During these workshops they developed local standards for emergency obstetric complications and women-friendly care based on the World Health Organisation (WHO) manuals and national guidelines. Common standards were developed for the three districts by health professionals working in maternity units and hospital administrators. Each hospital then measure its current practice, analysed data to identify gaps between current practice and standards, made recommendations for change and re-measured the practice after 3–6 months. Standards were developed for obstetric haemorrhage, obstructed labour, postpartum sepsis, pre-eclampsia/eclampsia, neonatal care, Caesarean section, and women-friendly care. Hospitals audited standards for emergency obstetric complications while health centres audited standards for women-friendly care because it was noted that emergency obstetric complications were rare in health centres, and when they did occur, they were often referred to hospitals.

Definition of terms 

Table 1, Table 2 present the definitions of terms used in this paper. Emergency obstetric complications refer to the nine complications: antepartum haemorrhage, postpartum haemorrhage, complications of abortion, pre-eclampsia/eclampsia, ectopic pregnancy, prolonged/obstructed labour, ruptured uterus, postpartum sepsis, and retained placenta. Emergency obstetric care was classified as basic (BEmOC) with six signal functions or comprehensive (CEmOC) with eight signal functions (Table 1). Population-based Caesarean section rate referred to Caesarean section as a proportion of the expected number of deliveries per year. Population-based Caesarean section rate was therefore based on estimates and not on actual figures obtained from the three districts. Expected number of deliveries per year was calculated as 5% of the population based on the national birth rate of 50.0 per 1000 population. The case fatality was defined as the proportion of women with emergency obstetric complications admitted to facility who die. Met need for EmOC was defined as proportion of women with emergency obstetric complications delivered at EmOC facilities. Facility-based maternal mortality rate was defined as the number of maternal deaths in health facilities per 100,000 deliveries.

Table 1. Definitions of obstetric emergency complications and obstetric signal functions

	Obstetric emergency complications (EOC)	Emergency obstetric care (EmOC) signal functions
	• Antepartum haemorrhage	Basic EmOC (BEmOC)
	• Postpartum haemorrhage	1. Injectable oxytocic drugs
	• Prolonged/obstructed labour	2. Injectable antibiotics
	• Puerperal sepsis	3. Injectable anticonvulsants
	• Pre-eclampsia/eclampsia	4. Manual removal of placenta
	• Ruptured uterus	5. Removal of retained products (e.g. MVA)
	• Abortion complications	6. Assisted vaginal delivery (e.g. vacuum extraction)
	• Ectopic pregnancy	Comprehensive EmOC (CEmOC)
	• Retained placenta	All the BEmOC signal functions 1–6 plus:
		7. Blood transfusion
		8. Caesarean section

Table 2. Definition of variables

	Variables	Definitions	Recommended level
	United Nation process indicators
	Availability of EmOC	Number of facilities providing EmOC per 500,000 population	1 CEmOC per 500 000 population; 4 BEmOC per 500 000
	Met need for EmOC	Proportion of women with obstetric complications delivered at EmOC facilities	100%
	Population-based Caesarean section rate	Caesarean deliveries as a proportion of all births (estimated from Crude Birth Rate)	5–15%
	Case fatality rate for emergency obstetric complications	Proportion of women with obstetric complications admitted to a facility who die	<1%
	Other indicators
	Institutional (facility) delivery rate	Proportion of all births in health facilities	–
	Expected number of deliveries per year	5% of the population (based on the Malawi Crude Birth rate of 50.0 per 1000 population)	–

EmOC: emergency obstetric care; BEmOC: basic emergency obstetric care; CEmOC: comprehensive emergency obstetric care.

Data collection 

Data were collected on the process and outcome indicators available in maternity registers. The health care providers working in maternity units were trained how to properly fill and interpret data recorded in maternity registers. Data was compiled from maternity registers on a monthly basis, by data clerks in hospitals and health care providers in health centres, and sent to their respective District Health Offices.

Statistical analysis 

Data on deliveries reported by 73 health facilities (13 hospitals and 60 health centres) in the three districts to District Health Offices over a 3-year period (41,637 deliveries in 2005, 43,729 deliveries in 2006, and 51,085 deliveries in 2007) were entered and analysed in SPSS version 15.0. The results are stratified where necessary and presented as absolute values or proportions (percentages). All reproductive health indicators were calculated based on the definitions in Table 1, Table 2.

Availability of EmOC services 

There were seven comprehensive emergency obstetric care (CEmOC) facilities in Lilongwe, one CEmOC facility in Kasungu and one CEmOC facility in Salima and this number did not change during the 3-year period (p for trend=1.000). The number of CEmOC facilities was adequate in two districts and slightly below the minimum recommended (1 per 500,000 population) in the third district. The distribution of CEmOC facilities was unequal with poor access in some rural areas.

There were 0.2 basic emergency obstetric care facilities (BEmOC) per 500,000 population (i.e. 1 BEmOC facility for an estimated number of 2,500,000 people) and this figure did not change during the 3 years (p for trend=1.000). With the exception of the functional CEmOC facilities, there was only one facility that provided all the six BEmOC signal functions during the 3 years of this study (2005–2007). The signal functions which require specific manual skills such as vacuum extractions and manual vacuum aspiration were the least available. This did not change over 3-year period (2005–2007). The details of the availability, utilisation and quality of EmOC services by district are presented in Table 3.

Table 3. Availability, utilisation and quality of emergency obstetric care (EmOC) services in three districts in Malawi over 3 years

	Variable	2005	2006	2007	p-value for trend
	Population (N)
	LL	1,720,784	1,871,958	1,951,278	–
	KS	589,019	608,917	650,103
	SL	308,882	331,308	342,979
	Total	2,618,685	2,812,183	2,944,360
	Expected deliveries (N)
	LL	86,039	93,598	97,564	–
	KS	29,451	30,446	32,505
	SL	15,444	16,565	17,149
	Total	130,934	140,609	147,218
	Facility delivery rate (%)
	LL	32.3%	30.0%	33.7%	<0.001
	KS	27.7%	27.6%	28.7%
	SL	37.1%	43.6%	51.6%
	Total	31.8%	31.1%	34.7%
	Maternal deaths (N)
	LL	66	66	55	<0.001
	KS	24	20	20
	SL	16	11	18
	Total	104	97	93
	Case fatality rate (%)
	LL	1.7%	1.6%	0.8%	<0.001
	KS	9.8%	4.9%	3.6%
	SL	3.1%	2.5%	1.7%
	Total	3.7%	3.0%	1.5%
	CEmOC per 500,000
	LL	2.0	1.9	1.8	1.000
	KS	0.8	0.8	0.8
	SL	1.6	1.5	1.5
	Total	1.7	1.6	1.5
	BEmOC per 500,000
	LL	0.3	0.3	0.3	1.000
	KS	0.0	0.0	0.0
	SL	0.0	0.0	0.0
	Total	0.2	0.2	0.2
	Met need for EmOC (%)
	LL	15%	17%	17%	<0.001
	KS	8%	12%	16%
	SL	30%	26%	34%
	Total	15.2%	17.0%	18.8%
	Caesarean section rate (%)a
	LL	1.4%	1.6%	1.4%	0.257
	KS	1.7%	1.2%	1.2%
	SL	3.1%	2.1%	2.6%
	Total	1.7%	1.6%	1.5%

LL: Lilongwe; KS: Kasungu; SL: Salima; EmOC: emergency obstetric care; BEmOC: basic emergency obstetric care; CEmOC: comprehensive emergency obstetric care.

a Population-based Caesarean section estimated as the number of Caesarean deliveries as a proportion of expected number of deliveries per year (estimated from Crude Birth Rate).

Utilisation of EmOC services 

Institutional delivery rate increased significantly over 3 years – 31.8% in 2005, 31.1% in 2006 and 34.7% in 2007 (p for trend<0.001). The population-based Caesarean section rate was generally low and decreased slightly over the 3 years, although the decrease was not significantly significant – 1.7% in 2005, 1.6% in 2006 and 1.5% in 2007 (p for trend=0.257). There was a significant increase in the met need for EmOC over 3 years – 15.2% in 2005, 17.0% in 2006 and 18.8% in 2007 (p for trend<0.001).

Quality of EmOC services 

Quality of care was assessed using mortality data and case fatality rate for emergency obstetric complications.15 The number of maternal deaths decreased monotonically (i.e. continuously) over 3 years and there was a statistically significant trend in facility-based maternal mortality rates – 250 per 100,000 women in 2005, 222 in 2006 and 182 in 2007 (p for trend<0.001). The overall case fatality rate decreased significantly over 3 years – 3.7% for 2005, 3.0% for 2006 and 1.5% for 2007 (p for trend<0.001). Case fatality rates for obstetric haemorrhage, prolonged/obstructed labour, puerperal sepsis, and pre-eclampsia/eclampsia decreased significantly but there was no significant change in the case fatality rates for abortion and ectopic pregnancy over 3 years.

Fig. 1 presents the change in case fatality rate and Caesarean section rate over the 3 years.

View Large Image Download to PowerPoint

Figure 1. Case fatality decreased while Caesarean section rate did not change over 3 years.