In Bangladesh formal health services are provided by a range of public, private and non-government organization (NGO) providers. It has been estimated that there are approximately 4,400 of first level government facilities in Bangladesh 11 .

The next tier of service is composed of some 417 sub-district hospitals called Upazila Health Complexes, which are designed to receive referrals from first level services. These usually have around 31–50 beds and are staffed by medical doctors, nurses, paramedics and laboratory technicians. There are also an estimated 59 district hospitals across the country each having between 50–250 beds 11 .

The WHO framework for improving hospital care to children under five years of age outlines seven interconnected steps in the quality improvement process 12 13 . These include: 1) orientation to the specific country and health system context, and the adaptation of tools for assessment; 2) undertaking a baseline hospital assessment; 3) using this assessment to facilitate a consensus on hospital care and support system standards; 4) using standards to inform the development of appropriate interventions and improvement strategies; 5) implementing interventions and improvement actions; 6) monitoring and evaluating improvement actions; 7) sharing information about hospital improvement and successes to inform fine-tune existing strategies, develop new ones and scale-up hospital improvement 12 .

The WHO has developed a number of generic assessment tools for evaluating the quality of hospital care to children; each of which has been developed based upon available evidence. The WHO Assessment Tool for Hospital Care for Children provides a set of standards for each of the most important dimensions of paediatric hospital care, and includes criteria for the assessment of performance in relation to each of these standards 12 .

Given concerns about the quality of, and limited information about hospital care in Bangladesh, we undertook a baseline assessment of the quality of care for under five children in randomly selected government district and sub-district hospitals in Bangladesh as these are where severely ill children are most likely to be referred to. We also sought to compare the current practices in these hospitals with evidence-based guidelines, using the WHO developed pocket book of hospital care for children, to initiate quality improvement in the health sector 14 . In so doing we hoped to identify priority areas in need of improvement and to provide recommendations on strategies that could assist in this regard.

Six district hospitals from a number of districts across the country were randomly selected for inclusion in the assessment. All the districts in each of the six divisions in Bangladesh were assigned a number. A random number generator computer program was then used to randomly select one district from each division, and then a district hospital was chosen from each of the six randomly selected districts.

Twelve sub-district hospitals were also included in the assessment. Sub-district hospitals in the selected six districts were stratified in two strata- hospitals having IMCI program and hospitals having both IMCI and emergency obstetric (EmOC) program. One hospital from each of the strata was selected randomly, using the same procedure applied for district hospitals selection (Figure 1).

Adapted versions of the WHO developed generic assessment tools were used for our assessment. Adaptation of the tools was conducted through workshops involving leading paediatricians, hospital administrators and relevant government and NGO programme personnel in Bangladesh. The tool was then field tested prior to use in the study.

The final assessment tool included sections on hospital layout, essential drugs, supplies and equipment; laboratory support; emergency area and emergency management; neonatal care and case management for acute respiratory infections, diarrhoea, febrile illness and malnutrition; supportive care and monitoring; hospital administration; access to hospital care; and quality of care.

A total of 31 assessors were selected to undertake hospital assessments. Assessors were physicians and GoB and NGO programme personnel with experience in the area of child health. All assessors received three days training, facilitated by individuals who participated in the initial adaptation workshop. Assessors also participated in field practice sessions.

Each team consisting of two paediatricians and two programme personnel conducted assessments at selected hospitals in May 2009. Assessments of each hospital took two days for teams to complete. Direct observation of clinical case management was undertaken wherever possible. When there were not sufficient patients for direct case observations, assessors reviewed clinical records. Assessors also conducted interviews with hospital staff and, if needed, used simulated cases when no case could be observed nor record could be found. The extraction of data from hospital Management Information System (MIS) reports was also used as another source of information.

Assessors scored hospitals on many of the dimensions in the adapted tool with respect to current WHO guidelines and standards, which were detailed in the tool. For individual and overall scoring, points from 5 to 1 were awarded: 5 for good practice complying with standards of care, 4 indicating little need for improvement to reach standard care, 3 meaning some need for improvement to reach standards of care, 2 indicating considerable need for improvement to reach standards of care and 1 for services not provided, or inadequate care and potentially life-threatening practices.

The results of completed hospital assessment tools were checked for consistency. A customized data entry programme was developed with the necessary consistency checks. After data entry and cleaning, the data was analyzed using SPSS version 12.

Approval for health facility assessment and data collection was obtained from Directorate General of Health Services under the Ministry of Health and Family Welfare of Government of Bangladesh. Hospital assessment was an activity which documented the existing practice and situation at public health facilities. Since the assessment of the hospitals was conducted as part of quality of care improvement process at government hospitals, not as a research, we did not obtain ethical approval from ethics committee.

Routine hospital MIS data showed that district hospitals (DHs) generally saw more children (M=17,435) than sub-district hospitals (SDHs) (M=8,728) in 2008. The proportion of children seen in emergency departments was greater in DHs (8.9%) as compared to SDHs (3.8%). A greater proportion of children were seen in outpatient departments in SDHs (86.3%) as compared to DHs (DHs: 80.8%), while the proportion of children seen in inpatient departments was similar (DHs: 10.3% and SDHs: 9.9%). The breakdown of hospital utilization by illness illustrated similar patterns for both district and sub-district hospitals (Figure 2). The most common illnesses in district hospitals (DHs) and sub-district hospitals (SDHs) were acute respiratory infections (DHs: 46% and SDHs: 52%), diarrhoea (DHs: 15% and SDHs: 18%), and fever (DHs: 6% and SDHs: 15%). Neonatal illness accounted for 6% of admissions in DHs but was negligible in SDHs.

All of the six DHs had a separate paediatric section and designated beds for children whereas seven SDHs had a separate paediatric section and only two had designated beds for children. None of the hospitals had paediatric surgeons. In terms of inpatient and outpatient care all DHs and SDHs had at least one doctor present during the morning shift (8 am-2 pm) but only a third of DHs (2 of 6) and two third of SDHs (8 of 12) hospitals had at least one doctor during the evening shift (2 pm-8 pm) and the night shift (8 pm-8 am). No DHs or SDHs had more than one doctor present in inpatient and outpatient departments during the evening and night shifts. All facilities had 24-hour nursing care for admitted children, and reported that emergency care provided by paramedics and/or on call doctors was available 24-hours a day and seven days a week. However, no hospitals had a functioning triage system for prioritizing severely ill children upon arrival at emergency departments (Table 1).

Electricity was available at all hospitals, but amidst the back drop of frequent power cuts in Bangladesh, only one DH and four SDHs had a backup generator. Sufficient fuel to power backup generators was not available at any SDH. Whilst running water was available at all but one hospital, more than half (10 of 18 hospitals) were unable to provide safe drinking water. Five DHs and eight SDHs had coloured bins for waste disposal but instructions for correct waste disposal were not always followed (Table 2).

Functioning refrigerators for Expanded Programme on Immunization (EPI) vaccines were present at all hospitals. Filled oxygen cylinders were available but three SDHs lacked flow meters. Half of DHs (3 of 6) and a quarter of SDHs (3 of 12) had incubators for neonatal care, and only one hospital had a warmer. Weighing scales for children were available at the inpatient departments of almost all DHs (5 of 6) and in just over half of SDHs (7 of 12) (Table 3).

Only one DH had all of the essential paediatric inpatient drugs (normal saline, cholera saline, injectable ampicillin, injectable gentamicin, syrup amoxicillin and injectable amoxicillin) whilst not a single SDH complied fully with the requirement. The majority of DHs (4 of 6) had a supply of normal saline, but this was present in less than half of SDHs (5 of 12). Cholera saline was found in almost all DHs and SDHs (5 and 11, respectively). Although both ampicillin and gentamicin should be available for the management of severe pneumonia or severe disease, at least one injectable antibiotic was available at all DHs, but not at all SDHs. Oral rehydration solution (ORS) and zinc tablets were available at all facilities (Table 4) .

Supplies for essential laboratory tests, such as blood glucose, haemoglobin, complete blood count (CBC), blood grouping and cross-matching, routine microscopic examination for urine and stool and facilities for chest x-rays were present in most DHs and some SDHs. However, the presence of laboratory test was undermined by the availability of laboratory support, particularly in SDHs. For instance, laboratory test results were available on a 24/7 basis in just a quarter of SDHs (3 of 12). Similarly half of the DHs (3 of 6) and one third of the SDHs (4 of 12) were not conducting laboratory tests according to standard operating procedures, and maintenance of laboratory safety measures was generally poor (Table 5).

Neither DHs nor SDHs had a separate unit or room for managing sick newborns. The safety of wards and hygiene practices relating to neonatal care was found to be poor overall. In particular substantial improvement was required in over half (7 of 12) of the SDHs in our sample. Likewise, no hospitals had adequate systems for prioritizing seriously ill neonates.

SDHs were found to be marginally better than DHs in promoting early initiation and exclusive breastfeeding and skin-to-skin contact for newborns (4 of 12 SDHs compared to 1 of 6 DHs). Adequate thermal protection for newborns was absent at all facilities and immunizations were not given routinely to all children born at the facilities.

Low birth weight (LBW) newborns were appropriately diagnosed at only two DHs. Treatment of LBW was poor at all hospitals, and particularly at SDHs, eight of which required substantial improvement compared to one DH. Similarly, appropriate feeding of young infants and LBW newborns was lacking at all hospitals (Table 6).

Prevalent child illnesses such as pneumonia, diarrhoea and severe malnutrition were poorly managed at these facilities. This was despite the fact that all hospitals with the exception of one, had workers trained in IMCI. Even so, our assessment found that there was very little retraining, nor adequate and updated guidelines available at hospitals.

Severity of pneumonia was found to be correctly assessed and diagnosed in only one-third of DHs (2 of 6) and in one-quarter of SDHs (3 of 12). Appropriate antibiotics were found to be administered inadequately in all hospitals. Similarly correct and appropriate administration of oxygen was absent in most of the hospitals (Table 7).

Diagnosis and treatment of diarrhoea was also generally found to be deficient in most hospitals. Substantial improvement with respect to this was particularly required in SDHs (8 of 12) as compared to DHs (2 of 6). Despite two DHs and two SDHs meeting standards, continued feeding during diarrhoea was found to require improvement in most hospitals (Table 7).

All hospitals fell below standards in relation to severe malnutrition case management. Appropriate nutrition assessment for malnutrition was not conducted in many hospitals. As a result two thirds of DHs (4 of 6) and three quarters of SDHs (9 of 12) were found to need substantial improvement in managing electrolyte imbalance and micronutrients appropriately. Similarly correct management of dehydration, hypoglycaemia and hypothermia, and the correct feeding of severely malnourished children were lacking (Table 7).

Re-assessment of all admitted children by nurses was not conducted in any hospitals. A comprehensive monitoring chart has been developed by the Government of Bangladesh, which describes patient details, vital signs, clinical signs, treatments given, feeding practices, an intake-and-output chart, and outcomes of these practices. It however, was not used at most of the facilities (Table 8).