Around one million women give birth annually in Iran, with 90% receiving maternity care in hospital. Maternal mortality is still high compared with rates in developed countries (37.5 per 100000 live births), as is neonatal mortality (16.9 per 1000 live births), and the caesarean birth rate is close to 40% 1. Of the 295 registered maternal deaths in 2005, 87.6% occurred in hospitals and 60% were found to be related to medical errors 2. Despite many advances in the Iranian health care system over recent decades, these statistics alone show that there is still much room for improvement in the quality of maternity care 3.

Worldwide, several methods have been used for improving the quality of maternity care in hospitals, such as clinical practice guidelines (CPGs), clinical pathways, and clinical audit 4567891011. Almost all of these methods have their origin in evidence-based practice (EBP) to apply the best evidence in clinical care 12. EBP is simply the integration of the best available research-based evidence, clinical expertise, and patient needs, values and preferences to develop a system of quality care 13. Quality improvements thus require professional consensus about implementation of research-based clinical evidence, and attention to patient needs, values and preferences 14.

Although EB practice has been recognised as necessary for quality care in developed countries, it has often been conceptualised in terms of research-based clinical evidence and less attention has been paid to patient needs and preferences 15. Yet service quality has been defined as meeting or exceeding service users' needs 1617. Patient satisfaction has indeed been recognised as an important outcome measure for the quality of health care since the late 1980s 18. Furthermore identifying patient needs and requirements has been judged essential for both measuring and improving quality of care 19.

In order to improve the quality of maternity care at an Iranian Social Security Hospital serving a poor population (Fayazbakhsh Hospital), we measured the impact on women's satisfaction and caesarean birth rates of a multifaceted intervention to improve uptake of selected EB practices, utilising a before-and-after study design.

Ethics approval to conduct the study was granted from the Ethics Committee of Tehran University of Medical Sciences and Urmia University of Medical Sciences (dated 20 September 2005) after submitting the information that would be given to the participants. All participants gave their written consent prior to the interviews and before-and-after surveys.

A four-step process was utilised to develop, implement and evaluate more evidence based maternity care at the study hospital 20.

Implementation of the new maternity care model improved clinician compliance with evidence-based guidelines and was associated with an improvement in women's satisfaction levels and a reduction in rates of caesarean birth. Whilst promising, these findings need to be interpreted cautiously primarily because of limitations for attributing causality in the before-and-after design used here 29. Thus, although our findings are consistent with those of other recently published studies 153031, the impact of the intervention would be more appropriately studied in a well-designed randomised trial. Whether the method and findings of the study have value in other settings and with other clinical processes will only be seen in future studies.

Most of the quality improvement literature emphasises the involvement of people who know the process best (the process owners) as essential for eliciting the best information and for increasing the participation and commitment of quality improvement team members 32. However there are some limitations. A dual role as caregiver and quality improvement team member can be a source of bias. In Step 1, the purpose of our interviews with women was to develop understanding of patient needs and requirements (not to assess their satisfaction with care). At this stage caregiver involvement was unlikely to introduce bias. However, the involvement of the midwives during the collection of satisfaction data (in Steps 2 and 4) may have been a source of bias in the study. We tried to increase the internal validity of the study through midwife training, assurances to women about the purpose of the study, standardisation of the survey process and supervision of the survey and data analysis by the authors who were external to the quality improvement processes 29.

The identified women's needs and requirements had a major role in design of the new care model. Available EB guidelines alone do not guarantee patient satisfaction or high quality of care. In this study women identified that they did not want support during labour and birth from their partner or a family member, but rather from their professional caregivers. Despite research evidence that supports the role for companionship during birth 33, women's stated preference was for non-professional support from their partners or family members, immediately after the birth. Thus the new model adopted a modified evidence-based guideline appropriate to the expressed wishes of the women attending the study hospital. In addition, professional consensus about areas of practice on which to focus improvement efforts largely ensured adherence to and smooth implementation of the new model of care. Our experience suggests that care designers should always consider all the relevant sources of evidence, including patient needs and requirements and professional consensus, in any EB initiative 34.

It appears that participation and raised awareness among clinicians played an important role in successful implementation of the new care guidelines. However, in around one in five cases compliance with the new model was not achieved. Changing care practices is always a time-consuming and ongoing process requiring much organisational support 35. Lack of time and experience, conflicts between professional groups and/or generations, cultural and psychosocial characteristics of the community, the physical and technical environment, rights and rules, and fear of poor outcomes may all be barriers to the adoption of EB practices 36.

In our study it was of interest that caesarean birth was more likely to be associated with non-compliance with admission guidelines, that is, women being admitted who were not in active labour, did not have effective contractions and did not respond well to stimulation – a set of circumstances that may predispose to a decision to perform a caesarean. Alternatively, for women who attended hospital late in labour there may simply have been less opportunity for medical intervention.

This multi-faceted intervention to translate appropriate evidence into practice appears to have had an important and positive impact on maternity care provision in the study hospital. Our study has demonstrated both that women are well aware of their needs and that identifying patient needs and requirements can play a major role in designing and measuring quality improvements. Equally, attention to professional consensus about priorities for EB practices appears to improve implementation of new care models and reduce clinician resistance to change. This study may also provide food for thought for health care policy-makers and care providers who are looking for better strategies to bring about evidence-based and patient-centred care.

The authors declare that they have no competing interests.

SA developed study design, administered the study team at the study hospital, performed data management and interpretation, and drafted the manuscript for publication. FA, AL, KM, RS and MA provided scientific advice on the design and implementation of the study, analysis and interpretation of the data as well as in drafting the manuscript. All authors read and approved the manuscript.