Widespread non‐adherence to evidence‐based maternity care guidelines: a population‐based cluster randomised household survey

To assess the quality of maternity care in an Indian metropolitan city.


Introduction
Quality of maternity care may be defined as 'the degree to which maternal health services for individuals and populations increase the likelihood of timely and appropriate treatment, for the purpose of achieving desired outcomes that are both consistent with current professional knowledge and uphold basic reproductive right'. 1 From the public health perspective this includes accessibility, availability and ability of healthcare resources to meet minimal care guidelines across the community. However, from a care-provider perspective this primarily denotes compliance with evidence-based guidelines, as is the case with medical conditions like diabetes. 2 Internationally, several studies have documented the quality of maternity care in both the public health and care provider contexts. [3][4][5][6][7][8] From the public health perspective these studies have served as independent audits of prevailing maternity healthcare systems. From the clinician perspective they have offered insight into the prevalence of non-evidence based practices, providing opportunities for self-improvement. In India, the National Family Health Survey (NFHS) and the District Level Household Survey (DLHS) have attempted to document the quality of maternity care at the population level. 9,10 However, these surveys have restricted themselves to an audit of minimal healthcare objectives such as three antenatal visits, universal iron supplementation and tetanus immunisation. Although these surveys provide useful public health data, they do not provide information on compliance with evidence-based guidelines. [11][12][13][14][15][16][17] There is an absence of universally agreed parameters for assessing quality of maternity care (though such measures exist for other conditions 2 ) and a paucity of literature on this subject internationally and from India. 18 This has precluded introspection at the health system or practitioner levels. Considering that quality of care is especially important in urban centres such as Delhi where accessibility of healthcare resources is less of a limitation, we conducted a pilot study in 2008 in South Delhi. 19 Our findings indicated that a substantial section of the populace were subject to overuse of investigations and interventions, while at the same time some of the minimal public health goals were not being met. The small study from a single locality did not allow generalisability or interpretation of the clinical-demographic determinants of the care provided. Hence we conducted the current 'DELhi DELivery CARE (DELCARE)' survey with the objective of evaluating the quality of maternity care in Delhi in terms of the degree to which common practices are evidence-based.

Maternity services in Delhi
Maternity care in Delhi, a city with a population of over 16 million, 20 is provided in a mix of public and private facilities that vary widely in infrastructure and staffing. This includes 32 public hospitals, 564 private hospitals and nursing homes, seven primary health centres and over 250 public maternity homes. 21 Most of the deliveries in the city are conducted by obstetricians. 20 The obstetricians in the private sector are typically in fee-for-service solo practices, whereas midwifery is poorly developed. 22 Requirement for reporting on outcomes related to maternity care is minimal at the health facility or practitioner levels. 23 Insurance coverage is limited and public facilities are often overburdened. 23

Subjects and methods
Participants were identified via three-stage cluster randomised sampling. Inclusion criteria were recently delivered women (180 days) who had a live, viable birth (after 28 weeks of gestation, confirmed by the primary care provider). Exclusion criteria were the inability of a participant to complete a questionnaire (subject unable to communicate, seriously ill, physical/mental disability or major cardiac, renal, hepatic, intestinal or neurological disease which required continuing treatment or hospital admission for >1 week within last year) or delivery outside Delhi.

Study design
The study design is presented in Figure 1. Delhi is divided into 150 wards (administrative subunits in the city, each electing one councillor; population 50 000-250 000 24 ). In stage 1, 20 wards were selected by a probability-proportionate-to-size systematic method, based on census data, using a random computer-generated seed value. 25 In stage 2, three colonies, one from each of three socio-economic strata, were chosen from each of the 20 wards (a colony is a small administrative unit usually 1-2 km 2 ; this constituted a cluster in our survey) by simple random sampling. In stage 3, a house-to-house survey was initiated in the chosen colonies. The starting point for each colony was the northern, southern, western or eastern extreme of the cluster (one of four was chosen at random), and the survey team moved lane-by-lane in the opposite direction (north to south, south to north, west to east or east to west) until all households were covered or a minimum of 30 willing participants were located. Permission of the local Resident Welfare Association (RWA) was sought wherever required. Participants were given a time for questionnaire administration within 2 weeks of the initial visit. The survey was conducted between September 2009 and February 2011. The cluster randomised approach stratified by socio-economic class was chosen to minimise bias and maximise representativeness and generalisability. Health facilities were classified into Hospitals (>25 beds; Public or Private), Nursing Homes (5-25 beds; private) and other small Institutions (<5 beds; public or private). Non-profit or NGO hospitals were included in the classification of private. Written informed consent was obtained. A free haemoglobin (Hb) test was done, and weight and height measurements were taken; there were no other incentives. The institutional ethics committee of the Sitaram Bhartia Institute of Science and Research approved the project.
Information was recorded on a standard peer-reviewed, pilot-trialled 19 questionnaire by patient recall and document verification wherever possible. This included maternal age, infant age, ethnicity, education level, annual household income, occupation, employment status and the place and type of health care provider. Socio-economic status was classified into higher, middle and lower using the income, education and occupation of the subjects in accordance with the inflation-adjusted Kuppuswamy scale (KSEC) 26 (scores 3-10 were classified as Lower Socio-economic Class, LSEC; 11-15 as Middle Socio-economic Class, MSEC; and ≥16 as Higher Socio-economic Class, HSEC). Detailed information was collected on quality-of-maternity care using standardised questions (see online Supporting Information). Details based on recall (not verified from records) were cross-checked from the subjects' husband/mother-inlaw in the case of any discrepancies. The bilingual questionnaire (Hindi and English; see Supporting Information Appendix S1 for English version; translated and back-translated) was designed to provide information on outcomes of interest. These included minimal care objectives from the national Reproductive and Child Health (RCH) programme such as iron supplementation, tetanus immunisation, number of antenatal visits and trained attendance during delivery using relevant questions based on the NFHS-3 questionnaire. 10 We also evaluated common care provider practices in comparison with evidence-based guidelines (mobilisation in labour, early initiation of breastfeeding, labour support, antenatal counselling for preterm labour and labour analgesia, etc. [11][12][13][14][15][16][17] and potentially overused or misused investigations and interventions such as induction, episiotomy, ultrasound, catheterisation, IV fluids, shaving of pubic hair. 17 For Hemocue haemoglobin estimation, 20 ll of blood was drawn by pinprick from the subjects to determine the prevalence of anaemia in the postpartum period. 27 Weight, height and haemoglobin estimations were performed at the woman's house.

Sample size considerations
On the basis of the pilot survey 19 it was estimated that a sample size of 1594 subjects would be required to calculate the prevalence of common physician practices (those with prevalence >10%) with an acceptable relative error of 10%. This sample size calculation accounted for the estimated number of deliveries per year in Delhi and an estimated design effect of 2.0 (accounting for the cluster design 19 ). We decided to recruit 30 participants from each of the 60 clusters from three strata based on the methodology adopted for the pilot survey.

Analysis
Data entry and analysis was done using PASW v 17.0 Software (SPSS Inc., Chicago, IL, USA). The results were adjusted for the three-stage stratified cluster design of the survey. Inter-and intra-cluster variation and the population of the cluster were used to 'weight' the results from the clusters, leading to an overall summary estimate of 60 such selected clusters. Complex Samples 'Descriptives' and 'Frequencies' procedures in PASW were used to provide adjusted means and frequencies where applicable. Differences between groups were evaluated on the basis of non-overlapping confidence intervals in estimates. Any missing data were treated as 'missing' and the data point was excluded from analysis but the subject retained.

Results
The survey team visited a total of 118 446 households (60 colonies, 20 each from the higher, middle and lower income groups; 344-6759 households per colony) and identified 2286 women who had delivered within the last 6 months. There were 185 exclusions as some women had delivered outside Delhi, had a 'major medical illness' or were 'unable to communicate'. In all, 300 women (14.2%) did not consent or were not traceable after three sequential visits to their residence. Hence, the questionnaire was administered to 1801 eligible consenting participants (588, 585 and 628 from the higher, middle and lower income areas, respectively). In all, 746 women were categorised as HSEC (41.4%), 233 women as MSEC (12.9%) and 813 as LSEC (45.1%). KSEC score could not be calculated for nine mothers as available information was incomplete/not given. Complete medical records (antenatal card, delivery papers and newborn discharge) were available for 998 participants (~55%) and one or more records were available for 1457 participants.
The socio-economic and obstetric profiles of the participants are shown in Table 1. The average age of the mothers varied between 25.0 and 28.2 years across the socio-economic spectrum. One-third of mothers were primiparous with a trend towards higher parity in lower SEC. The proportion of anaemic women was higher in the lower SEC (47.2%) than the higher SEC (31.7%). An obstetrician provided antenatal care to >90% mothers from the middle and higher SEC in comparison with 66% in the LSEC. More than half of the deliveries took place in hospitals. Private hospitals and nursing homes conducted 81.2% of deliveries in the HSEC while public hospitals conducted 16.8%. In the LSEC, 16.3% of all deliveries took place at private hospitals and nursing homes, whereas public hospitals conducted 50.2%; 28.8% of LSEC deliveries were conducted at home.
The obstetric risk profile of the women was compared across the healthcare facilities for maternal age, primiparity, non-cephalic presentation, low birthweight (LBW; <2500 g), prematurity and twinning. There was no significant difference in the incidence of teenage pregnancies, pregnancy in ≥35-year-olds, premature deliveries and non-cephalic presentations. Low birthweight babies were born to 22.8% mothers at public hospitals compared with 14.1% at private hospitals and 17.9% at nursing homes. Of mothers delivering at private facilities, 54.4% were primiparous compared with 42.2% at public hospitals.
Minimal national goals such as iron supplementation advice, tetanus vaccination and at least three antenatal visits were met in >80% of the population, with the exception of those delivering at home. Women in private hospitals had a significantly higher number of ultrasounds (mean 4.3 versus 2.5), and significantly higher caesarean (53.8 versus 23.7%) and induction (30.8 versus 20.6%) rates compared with public hospitals. Private facilities, especially private hospitals, fared better at counselling aspects: preterm labour counselling was provided to 15.0 versus 44.2% and postnatal exercise counselling to 8.5 versus 28.3% of women in public and private hospitals, respectively. Auspicious timing was self-reported as the reason for induction by 3% of mothers delivering at private facilities. Only~1% of women received labour support and pain relief in labour in public hospitals. Episiotomy rates were universally high, and antiquated practices such as routine enemas, urinary catherisation and shaving of pubic hair were widely prevalent (7-60% across facilities, excepting home deliveries). Family planning advice at discharge was given to only a quarter of mothers with no significant differences according to healthcare facility.
The quality of neonatal practice is summarised in Table 4. While a significantly higher proportion of the deliveries at private facilities were attended by a child specialist, non-ideal norms such as routine shifting of the baby to the nursery and delayed initiation of breastfeeding continued to be widely prevalent. Introduction of formula/animal milk feeding within 48 hours of birth and higher rates of phototherapy were also reported in private facilities. A minority of the babies stayed with the mother in the first hour after birth across the health care spectrum.

Main findings
This cluster randomised, community-based survey reveals a poor quality of a range of maternity care practices across the healthcare spectrum in Delhi, India, and points to systemic determinants of quality of care. Although some basic public health objectives are being met, prevalent practices deviate substantially from a wide range of national and international evidence-based guidelines. This includes private institutions where the patient or insurance provider pays for care and where infrastructure resources may be less limiting. The survey highlights the overuse of investigations (ultrasounds) and interventions (caesarean section, induction, and episiotomy), especially in private facilities, and a deficiency of patient-centred practices (such as counselling and labour support) particularly in the public hospitals.

Strengths and limitations
The strengths of the study lie in sampling strategy, a large sample size, a patient-centred approach and the comparison of community-based data with evidence-based guidelines. The collection of data by all-women staff using a validated (pilot trialled and peer reviewed) bi-lingual (Hindi and English) questionnaire also enhances the credibility of the results. The study covers a socially, culturally, educationally and ethnically diverse urban population with a multiplicity of health care models and with patchy penetrance of insurance coverage. While poor infrastructure and affordability continue to be cited as the major reasons for limited progress in maternity care across the country, 28 the current study presents data from a city where many of these factors may be less significant limitations, shifting the focus to the healthcare providers and systematic flaws (Delhi has one of the highest per-capita bed capacities and per-capita income in the country 21 ). The questionnaire-based survey methodology is limited by biases of recall, recency, belief and hindsight. However, lack of reliable birth registration data and the amorphous multiplicity of service providers precluded any other study design. The quality of care may have been overestimated due to the exclusion of non-live births. The  cross-sectional nature of the information limits any causeeffect inferences. The survey represents an urban metropolis which limits generalisability to the largely rural countryside but may be comparable to similar settings across the developing world.

Interpretation
Non-adherence to evidence-based standards has been reported for individual elements of our survey both globally 6,7,29,30 and within India 18,19,22 but we have found no comparable study examining a wide range of patient-centred outcomes relating to a multitude of evidence-based guidance. Studies from developed countries report a lower prevalence of outdated practices (such as pubic hair shaving, urinary catheterisation and routine enema 6,7 ), lower use of investigations (only 9.2% of women at the National Maternity Hospital in Ireland had >2 ultrasounds 31 ), and lower rates of interventions. 5 Our results are more comparable to those from Brazil, another developing country, which has high rates for interventions and low rates for patient-centred processes of care (for example, only 1% of deliveries have labour support). 32 Within India, national programmes such as the Reproductive and Child Health Programme 33 and National Population Policy 34 have identified a minimum of three antenatal visits, iron supplementation, and tetanus-toxoid vaccination as key outcomes; our results for these are similar to those reported in the NFHS-3 and DLHS for Delhi. 9,10 The caesarean rate found in our survey is much higher than the 17% rate reported by NFHS-3 for urban India but is comparable to the rates found in Chennai. 18 What our survey adds is information on other important aspects of maternity care, such as number of ultrasounds, labour support, labour analgesia and phototherapy. In line with another study from urban India, 35 we found high rates of discarding colostrum and delay in breastfeeding, in contravention to guidance from the Indian Academy of Paediatrics. 36    The widespread departure from evidence-based guidelines is of concern for several reasons, primarily the avoidable maternal and neonatal morbidity. Medical investigations should be carried out only when indicated and their indiscriminate use is known to increase false-positives. 37 This implies that the high number of ultrasounds and high electronic fetal monitoring rates may be contributing to the high caesarean rate, 38,39 which will in turn fuel an increase in related short-and long-term complications. 40,41 Excessive use of ultrasounds may also increase the risk for neurodevelopmental disorders. 42,43 The low rates for labour support and pain relief for the majority of women across facility types are inconsistent with international guidance. 14,17 It can be hypothesised that overcrowding and disempowerment of users contribute to some of the care practices seen in public hospitals. Low achievement of national health targets related to antenatal care in home deliveries probably reflects absence of care rather than a problem with the home setting per se. 44,45 The higher intervention rates at private facilities could be hypothesised to reflect fee-for-service financial incentives, time pressures of single obstetric practice, fear of litigation, patient preference, or widespread use of ultrasounds and electronic fetal monitoring without proper indications and understanding of risk-benefit analysis. The absence of midwifery 46 is probably an important limitation across facilities, as obstetrician-led practices worldwide have been shown to have higher interventions. 47  Minimal outcome reporting requirements allow the unsatisfactory situation to remain under the radar.

Conclusion
The DELCARE survey documents prevalent practices using a population-based, community survey that provides a wider 'diagnostic' view of maternity care in Delhi. It shows that care falls substantially short of evidence-based national and international guidelines. This information should sensitise policy makers and providers towards the need for better health governance. Further work should develop a uniform set of quality measures and registries for maternity care to enable comparisons across standards, institutions, geographic areas and time. Research should be undertaken to understand the reasons for such widespread departure from guidelines and to determine how care can be better aligned with evidence.

Disclosure of interest
None reported.

Contribution to authorship
JN, RSG, VLB and AB contributed to the concept and design of the study. AS was involved in data collection, checking, analysis and drafting of the manuscript. AB and JN finalised the manuscript. AB will act as guarantor for the project. All authors, external and internal, had full access to all of the data (including statistical reports and tables) in the study and can take responsibility for the integrity of the data and the accuracy of the data analysis.

Details of ethics approval
The project was approved by the institutional ethics committee of Sitaram Bhartia Institute of Science and Research (Dated 3/10/2008; SBISR/IEC/2008/03).

Funding
Supported by a grant from the Indian Council of Medical Research. The Indian Council of Medical Research approved the project protocol for funding but had no role in data collection, analysis or interpretation, in the writing of the report or in the decision to submit the article for publication. The researchers were completely independent from the funders.