Study design
This is a retrospective study that used secondary medical and financial data extracted
from Vietnam National Children’s Hospital and Thanh Hoa Provincial Pediatric Hospital.
Health system settings
Vietnam’s public healthcare system is divided into four administrative levels: The
first level is the central or national level (Level I); the second level is the provincial
level (Level II), the third level is district level (Level III), and the fourth level
is the communal level (Level IV).
Level I facilities include national general and specialized hospitals, national research
institutes, training institutions, pharmaceutical companies. These facilities are
mostly located in large cities and subordinate to the Ministry of Health (MOH). At
the central level, the MOH is also responsible for formulating and executing health
policies and programs for the entire country.
Level II facilities includes municipal and provincial hospitals, including general
hospitals and specialized hospitals such as pediatric hospitals, obstetrics and gynecology
hospitals, hospitals of ophthalmology, and other specialized health centers such as
preventive medicine centers and mother and child’s health protection centers.
Level III facilities include district general hospitals responsible for curative services
and preventive medicine defined in the national preventive program guidelines. Some
rural areas have one or more polyclinics that operate under the guidance of the district
hospital. These clinics mainly provide basic curative care for people living in communes
of the districts.
Level IV facilities mostly include Community Health Centers (CHCs) that are responsible
for primary curative and preventive care as well as implementation of national health
programs, including maternal and child healthcare programs. Within CHCs, village health
workers provide health information, education, and communication. They also provide
first aid and care for common diseases, implement family planning, and other national
health programs.
Under Vietnamese health insurance law (2009), all children under six years of age
are automatically covered by national health insurance. In principle, this insurance
covers all costs associated with premature births as long as parents seek treatment
at their assigned facilities, and do not seek specialized (hospital) treatment without
obtaining prior approval and referral from a basic health facility.
If a child is born prematurely, it is usually transferred from the delivery room to
the pediatric ward for intensive care right after birth. Depending on the child health
status, children may – upon consultation with the parents – be transferred to specialized
hospitals. In such circumstances, preterm infants born in a community health center
or district hospital are typically transferred to the provincial pediatric hospital.
From a provincial pediatric hospital, children with several severe health conditions
may be referred to the National Children’s Hospital.
Study setting
All data analyzed in this study were collected from two large hospitals in Vietnam.
Despite major efforts to reduce neonatal mortality, in 2015, 12 per 1000 live births
died during the neonatal period in Vietnam (10). Preterm birth complication have remained
a leading cause of neonatal deaths (41%, 2015); with 9% of all infants born before
37 weeks of gestation (11).
Two hospitals were purposively sampled for this study: Vietnam National Children’s
Hospital (VNCH) and Thanh Hoa Pediatric Hospital. Thanh Hoa Pediatric Hospital is
a provincial hospital (second level) and the only public pediatric hospital in Thanh
Hoa Province. As such, it is responsible for the treatment of all children with severe
health problems in the province. The province comprises 24 districts, 1 town, 2 cities,
and a total population of 3.5 million inhabitants (12). The area is environmentally
heterogeneous including lowlands, mountains, and seaside areas. It also has a whole
range of economic sectors such as agriculture, forestry, fishery, tourism, and an
industrial center (12).
Vietnam National Children’s Hospital is a national hospital (the first level) located
in Hanoi City. It is a primary referral hospital for all 38 provinces of northern
Vietnam, as well as a center for research, teaching, and postgraduate training in
newborn diseases. As the largest pediatric hospital in northern Vietnam, the Neonatal
Care Unit of the hospital is responsible for the treatment and care of all premature
infants and newborns referred from lower-level facilities for specialized care. This
covers a total population of 43.2 million people.
Participants
A preterm child in this study was defined as an infant born alive before 37 weeks
of gestation. We defined our target population as all infants discharged from the
two hospitals between 1 January and 31 December 2017.
Sample size and sampling method
The records for all preterm infants (n=261) discharged from the Neonatal Care Unit
of Thanh Hoa Pediatric Hospital in 2017 were selected. Due to the large number of
cases, a random sample of infants (n=500) was selected from all those discharged from
the Neonatal Care Unit at the National Children’s Hospital (n=800) during 2017. This
n=500 target was chosen in order to be able to detect a mean difference in total cost
between the two hospitals of at least 25% of the standard deviation of individual
total costs with 90% power.
Variables
Outcome variables: The primary outcome variable was the total numerical medical cost. All cost data
were extracted from the hospital records and converted to USD using an exchange rate
of 1 USD=23.245 Vietnamese Dong (VND) (2018).
Total direct medical costs were defined as the total amount invoiced by a hospital
after the child was discharged from the hospital. The total direct medical cost was
then divided into cost paid by national health insurance and the out-of-pocket cost
paid by parents.
Covariates: Social-demographic and clinical characteristics of premature infants were extracted
from the hospital records. These included: gender (male, female), ethnicity (Kinh,
others), place of residence (urban, rural), length of stay (<14 days, 15-29 days,
>30 days), weight at birth (>2500g, 2499-1500g, 1499-1000g, <1000g), gestational age
(32-37 weeks, 28-31 weeks, <28 weeks), place of referral (health facility, home),
transfer type (proper transfer, improper transfer), and referral type (from home to
Thanh Hoa Pediatric Hospital, from health facility to Thanh Hoa Pediatric Hospital,
from home to Vietnam National Children’s Hospital, and from health facility to Vietnam
National Children’s Hospital).
Statistical analysis
Firstly, detailed statistics on total direct medical cost as well as costs per service
category were generated. Secondly, we estimated the relative cost of being referred
to a national rather than a provincial hospital for treatment using a series of linear
regression models. We first quantified the average cost differences between the provincial
and the national hospital including and excluding child characteristics. The main
independent variable of interest was a dichotomous variable for the child being treated
at the national hospital, using the provincial hospital as a reference. In the adjusted
model, we included the following covariates to account for potential differences in
medical need: gender, ethnicity, place of residence, weight, and length of stay.
Thirdly, we further divided the sample into four types of basic health system trajectories:
1. Infants’ families seeking care directly (without prior referral from level III
or IV) at Thanh Hoa Pediatric Hospital; 2. Infants referred from a local health facility
to Thanh Hoa Pediatric Hospital; 3. Infants’ families seeking care directly at Vietnam
National Children’s Hospital; and 4. Infants referred from lower-level health facilities
to Vietnam National Children’s Hospital. Following this, we first estimate unadjusted
associations with cost (mean cost differences), and then estimated how large the cost
differences were adjusting on observable child characteristics. We used robust variance
estimates to adjust for heteroskedasticity of residuals in the linear regression models.
Given that appropriate referrals are critical for reimbursement of out-of-pocket payments
by the insurance, we also looked directly at the relationship between transfer type
and out-of-pocket expenditure in different referral groups. A transfer was considered
as a proper transfer to a higher-level hospital if the child had referral documents
from a local registered facility or was admitted to the hospital as an emergency case.
All analyses were performed using the STATA statistical software package (Release
14; College Station, TX: StataCorp LP).