Article Text
PDF
Abstract
Objective To estimate inpatient care costs of childhood severe pneumonia and its urban–rural cost variation, and to predict cost drivers.
Design The study was nested within a cluster randomised trial of childhood severe pneumonia management. Cost per episode of severe pneumonia was estimated from a healthcare provider perspective for children who received care from public inpatient facilities. A bottom-up micro-costing approach was applied and data collected using structured questionnaire and review of the patient record. Multivariate regression analysis determined cost predictors and sensitivity analysis explored robustness of cost parameters.
Setting Eight public inpatient care facilities from two districts of Bangladesh covering urban and rural areas.
Patients Children aged 2–59 months with WHO-classified severe pneumonia.
Results Data on 1252 enrolled children were analysed; 795 (64%) were male, 787 (63%) were infants and 59% from urban areas. Average length of stay (LoS) was 4.8 days (SD ±2.5) and mean cost per patient was US$48 (95% CI: US$46, US$49). Mean cost per patient was significantly greater for urban tertiary-level facilities compared with rural primary–secondary facilities (mean difference US$43; 95% CI: US$40, US$45). No cost variation was found relative to age, sex, malnutrition or hypoxaemia. Type of facility was the most important cost predictor. LoS and personnel costs were the most sensitive cost parameters.
Conclusion Healthcare provider cost of childhood severe pneumonia was substantial for urban located public health facilities that provided tertiary-level care. Thus, treatment availability at a lower-level facility at a rural location may help to reduce overall treatment costs.
- Child Health
- Child Health Services
- Health Care Economics and Organizations
- Health services research
Data availability statement
Data are available upon reasonable request.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Childhood severe pneumonia is the leading cause of under-5 mortality worldwide and majority of deaths concentrated in low-income and middle-income countries (LMICs).
Management of severe pneumonia requires inpatient care with careful monitoring that is associated with substantial resource usage.
Previous reviews summarised inpatient care costs for childhood severe pneumonia in LMICs, ranging from US$42 to US$1,818 per patient, with an average length-of-stay of 5.8 days.
WHAT THIS STUDY ADDS
The study estimated inpatient care cost per episode per patient (US$48) using patient-specific data from a sizeable sample (n=1252) in selected public healthcare facilities.
Identified common resource use, estimated per-patient cost in urban and rural settings with its distribution per cost component, and determined cost by level of healthcare facilities.
Comprehensively estimated cost per patient for each included facility (n=8), assessed important cost predictors and determined cost-sensitive parameters.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
The per-patient cost was found to be lower in rural facilities, but enrolment was higher in urban facilities in our sample, indicating a significant cost burden for the country’s health system.
Childhood severe pneumonia management should be made more accessible in primary-level facilities, which could potentially lower treatment costs by only referring complex cases to tertiary-level facilities.
Introduction
Childhood pneumonia is a significant cause of under-5 morbidity and mortality with an estimated 0.74 million deaths in 2019, concentrated in low-income and middle-income countries (LMICs).1 2 Of childhood pneumonia episodes, 16% progress toward severe pneumonia and contribute to the mortality burden.3 In Bangladesh, 14% of the approximately 0.15 million under-5 deaths every year are caused by pneumonia.4 Recent evidence reveals that care was sought for only 42% of children with pneumonia symptoms.5 Despite this, pneumonia accounts for about 40% of paediatric hospital admissions and 60% of outpatient visits in Bangladesh.6
Management of childhood severe pneumonia usually requires hospital management with careful monitoring as per WHO guidelines for LMICs,7 with associated resource use and costs for families (eg, out-of-pocket expenses) and healthcare providers (eg, inpatient care provision). These economic consequences impact the overall healthcare system meaning that the disease is an important health burden, particularly for resource-poor settings like Bangladesh.8 A review summarising economic evidence of childhood severe pneumonia revealed high inpatient care costs in LMICs (range: US$42–1818; 2022 prices) with an average length of stay (LoS) of 5.8 days.9 Reviews highlighted the limited economic evidence and methodological limitations of LMIC studies, with particular lack of detailed costing methods using patient-specific data for resource-constrained settings.9 10
Given the high burden of childhood severe pneumonia in Bangladesh, estimation of provider costs is vital to inform health system policymakers in order to optimally (re)allocate resources, to treat more patients within available resources. Several recent clinical trials conducted in Bangladesh reported brief cost information of inpatient care management; however, all were subjected to small sample size and covered a single urban facility.11–13
A large-scale clinical trial covering urban and rural areas was conducted by the International Centre for Diarrheal Disease Research of Bangladesh between 2016 and 2019 to assess the clinical effectiveness of a new daycare approach (DCA) over usual (inpatient) care management (ClinicalTrials.gov Registry: NCT02669654).14 Data from this trial enabled estimates of household’s cost for receiving inpatient care for childhood severe pneumonia; mean cost per household was US$147 and 65% of the cost was the productivity loss of caregivers.15 The current study aimed to supplement earlier household cost findings by estimating cost per case of childhood severe pneumonia from a healthcare provider perspective. Together, these studies present a comprehensive estimate of the economic burden of childhood severe pneumonia management in Bangladesh.
Method
Study setting, recruitment and sample
The trial was within two sites of an urban district (Dhaka) and two sites of a rural district (Kishoregonj). From these sites, 32 clusters were selected and randomised to either intervention (DCA) or control (usual care). Children aged 2–59 months with WHO-defined severe pneumonia16 with/without malnutrition were enrolled.14 Children were initially assessed in the cluster and then referred to local public or private inpatient care facilities. This analysis considered enrolled children from the control arm who received inpatient care from public facilities. Children from eight urban clusters received care from four urban-located facilities (tertiary-level hospitals); and children from eight rural clusters received care from four rural-located facilities (one secondary-level district hospital and three primary-level upazila (rural subdistricts) health complexes).14 Treatment costs at private facilities were assumed to be passed on in full as fees to households and are therefore fully captured in the household costs previously reported.15
Data collection
Trained and experienced interviewers conducted questionnaire-structured interviews with relevant personnel (eg, management and finance staff) from each included facility. Patient-specific cost data were collected from medical records of enrolled patients during their hospital stay, after discharge and during follow-up visits for up to 3 months. Patient demographic and socioeconomic information were collected through a separate structured questionnaire through interviewing parents/caregivers.
Costing method
A bottom-up micro-costing approach was applied using a healthcare provider perspective for identification, measurement and valuation of resources consumed in the provision of care.17 First, a list of all resources used in one episode of care was developed, including capital and recurrent cost items (eg, personnel, healthcare and non-healthcare consumables). Next, measurement of individual-use and shared-use items was undertaken. For example, we categorised staff and then determined staff time (in minutes) for each patient. For valuation, the unit cost for each listed resource was determined by its market price (eg, staff time was valued at the hourly wage) or estimated opportunity cost, for example., the equivalent annual cost for each capital item was calculated by allocating the purchase price across its estimated lifetime using a 3% discount rate.18 For instance, the presumed effective lifespans for electronic items/equipment, furniture and buildings were 5, 10 and 30 years, respectively.18–20 Full details of the costing approach for capital/recurrent cost items are described in online supplemental table 1. Per-patient costs were calculated by multiplying the unit cost of each item with the quantity consumed by each patient. Finally, patient-specific costs were obtained by summing up all relevant cost items used for each patient.
Supplemental material
Data analysis
All collected data were checked by field supervisors and researchers. Descriptive analysis presented average resource use and costs overall, by cost category (eg, medicines, personnel) and by potential cost driver variables (eg, type of health facility, age). Cost variation was measured using parametric or non-parametric statistical significance tests, as applicable. Variables that showed significant cost differences were included in an adjusted regression model. Data were skewed; therefore, logged-linear regression models were adopted to explore the predictors of log-transformed provider costs. Low value of variance inflation factor confirmed non-presence of multicollinearity. One-way sensitivity analysis was undertaken to explore how results varied from base case values with ±20% changes in cost parameters and ±1 day LoS. Statistical significance was considered as a p<0.05 and presented with 95% CIs. All analyses were performed using the statistical software STATA V.16.1 (Release 16. College Station, Texas, USA: StataCorp). Costs were expressed as US$ (US$1=84.5 Bangladeshi taka) in the 2019 price year (trial end year).
Results
Study participants
A total of 1252 children were enrolled, with a mean age of 12.6 months (SD ±10.5), and 64% children were male and 63% from urban areas (table 1). The average LoS in hospital was 4.8 days (SD ±2.5), with 55 patients (4.4%) referred to higher-level hospital(s), mostly from rural areas (n=51).
Background characteristics of enrolled children with severe pneumonia (n=1252)
Resource utilisation pattern
Healthcare resource used included physicians and nurses (100%) followed by medicines (86.2%) and nebulisation (76.9%) (table 2). The resource usage pattern was significantly different between urban-located and rural-located facilities, with greater use of diagnostic tests, nebulisation and oxygen in urban-located facilities.
Resource use pattern, overall costs, urban–rural cost distribution by each cost parameter
Average cost per patient with urban–rural cost distribution
Mean provider cost per patient per episode for childhood severe pneumonia was US$48.1 (95% CI: US$46.4, US$49.9) (table 2). Personnel costs (eg, physician and nurse) were the major cost contributors (US$23.7, 49%). Overhead costs were the second-highest cost component (US$9.7, 21%) followed by nebulisation cost (US$7.4, 15%).
Mean costs were significantly different between urban-located and rural-located health facilities (difference US$42.9, 95% CI US$40.3, US$45.7) and this difference was seen across most cost components (table 2). Per-patient cost was significantly higher for urban-located tertiary-level care compared with rural-located primary-level healthcare facilities (difference US$42.6, p<0.001) (figure 1). Despite both these components being accessed similarly in both urban and rural-located facilities, costs for physician, nurse and nebulisation were significantly higher in urban-located facilities (table 2).
Inpatient care mean cost variation (US$, 2019 price year) by facility types/levels for the management of childhood severe pneumonia. UHCs, upazila health complexes.
Significant cost variation was found in LoS categories in which mean cost per patient was greatest if LoS exceeded 6 days (US$59.9, 95% CI: US$51.3, US$63.3) (table 3). Overall, provider costs appeared to be not associated with patient age, sex and severity (presence of hypoxaemia and/or malnutrition). Complex cases (children with hypoxaemia and/or malnutrition) were mostly enrolled and treated in urban-located tertiary facilities with a significant higher cost compared with rural-located primary/secondary-level facilities (online supplemental figures 2–4).
Variation of provider cost by clinical characteristics (US$, 2019)
Mean healthcare provider cost and mean cost per parameter for each of the included healthcare facilities by facility location are presented in online supplemental tables 2 and 3. The mean cost per patient ranged from US$49.5 to US$80.1, and costs significantly varied across facilities (p<0.001). For rural facilities, mean provider cost per patient ranged from US$21.0 to US$23.2.
Predictors of healthcare provider costs and cost-sensitive parameters
The adjusted multivariable logged-linear regression model demonstrated that LoS, referral cases and level of facility were the significant predictors to determine provider cost. For example, an additional day’s LoS in hospital was associated with an 11% increase of log-transformed provider costs (p<0.001) and treatment at tertiary-level healthcare facility was associated with a 62% increase in mean healthcare provider costs compared with primary-level facility. No significant association was observed for clinical variables (table 4). The most sensitive cost parameters included LoS and personnel cost. For instance, a day increase in LoS or increased personnel costs by 20% increased per patient cost by 21% and 10%, respectively (online supplemental figure 1).
Adjusted multiple linear regression model with log-transformed provider costs (n=1252)
Discussion
This study estimated healthcare provider costs for inpatient care management of childhood severe pneumonia, presenting to urban or rural facilities from data collected through a large trial. The study expands the available literature, which is currently limited for an LMIC context. The mean cost per patient was US$49 and costs were substantially higher if treatment was provided in urban-located tertiary-level facilities. Cost varied significantly by level of facilities and by LoS.
In our study, personnel and medicines were the most common resources used for severe pneumonia management. Similar utilisation patterns are reported by studies conducted in other LMIC settings.18–21 Other resource categories did not appear as main items of provider costs in our study but are more common resource items for household costs.15 The mean healthcare provider costs are either comparable or lower than the estimates reported by other LMICs.9 Estimated mean provider costs (inflation-adjusted to 2019) were either similar or higher in Gambia (US$67),22 Kenya (US$191),19 Pakistan (US$302),23 Zambia (US$254 per bed-day)24 and Vietnam (US$286)18 compared with our observations. Variation in healthcare personnel salaries across different countries might be a potential reason for any dissimilarities in mean provider costs. Medicine cost is US$7 in our study, comparable with estimates reported for India (US$7)25 and lower than estimates in Pakistan (US$42).23 The higher medicine costs in Pakistan might be due to the high cost-sharing by public health systems where low medicine costs in our study reaffirmed high cost-sharing by households for required medicines.15 25
Estimated provider costs were much higher in urban-located higher-level facilities, similar to other LMIC studies. A study in India reported high mean per patient costs in tertiary-level hospitals (US$171) compared with secondary-level hospitals (US$98).25 The underlying reasons behind this might include high facility-level expenditure, such as building costs or rent, advanced equipment, potentially more resource availability as well as more staff time, variation in discharge policy and resource usage due to disease severity. For example, our findings indicate that most of the enrolled children with hypoxaemia and/or malnutrition received treatment at tertiary-level facilities, suggesting that more severe cases are often referred to and treated in higher-level facilities. Additionally, the high average LoS in urban facilities could be another factor that contributes to the higher cost burden because of more resource requirements to manage complex cases. Further research is recommended to explore the difference in LoS between urban and rural facilities.
In our study, diagnostic costs were much higher in urban tertiary-level facilities than rural facilities. Similarly, Madsen et al 25 showed diagnostic tests differences between secondary-level and tertiary-level hospitals in rural India settings (US$7 and US$26, respectively). However, cross-country comparisons of provider costs are limited because of different health service delivery systems, differences in unit costs of resources, treatment protocols and estimation approaches. In addition, high expenditure from households for the same cohort of patients15 explains the relatively low per-patient provider costs reported in our study compared with other LMICs.
LoS in hospital was the most important cost predictor and the most sensitive cost parameter in our study, which is consistent with recent reviews on pneumonia management.9 10 Per-patient provider cost varied across referral cases, with no cost variation in relation to other clinical variables such as age, sex and presence of hypoxaemia. Similar to our study, a study in Turkey found non-significant cost variation across age groups; however, in contrast to our findings, the study showed cost variation across gender with higher costs for boys.26
Overall, our findings reveal that mean childhood severe pneumonia treatment cost from a provider perspective is slightly lower in Bangladesh compared with similar settings, particularly for rural-located facilities. However, the government of Bangladesh covers only 23% of health expenditure that lucidly explains the relatively low provider costs since most of the costs have to be managed by households. For instance, cost from a household perspective using the same data revealed high costs (US$147) on average, especially when treatment was received from urban-located facilities (US$177).15 Evidence from available LMIC literature reveals that households in Bangladesh bear a relatively higher proportion of treatment costs compared with other countries.19 20 25 On the other hand, that 80% of the pneumonia-specific deaths occurred outside hospitals27 28 clearly indicates potential barriers to receiving healthcare services including service unavailability, unaffordability and high cost of care. In Bangladesh, one public hospital bed-day is allocated for more than 1600 people and most of the health workforce is urban centred.29–31 In rural areas, along with the shortage of skilled workforce, facilities suffer from retention and absenteeism of the public health workforce.30 The lack of sufficient health workforce might influence low care-seeking in rural areas and implies increased challenges of inpatient care service provision. The findings suggest that it is possible to allocate the same financial resources to treat more children if similar services and treatment quality can be provided in rural-located facilities to treat childhood severe pneumonia.
Strengths and limitations
Major strengths include the application of a micro-costing bottom-up approach using patient-level data to estimate cost per patient, informed by literature reviews.9 10 The analysis included a large sample and estimated detailed cost for each included facility covering urban and rural areas. Despite the strengths, the study has certain limitations. First, costs of some shared items were unavailable; therefore, mean cost from another facility was used as a proxy for those items. Second, for a small number of patients referred to higher-level facilities outside of the enrolled facilities of the project, we were unable to track costs of follow-up visits. Third, while the study covered both urban and rural locations, it was limited to two districts which may limit generalisability of findings beyond these districts. Finally, this study did not formally combine costs of severe pneumonia management from the provider perspective with household costs; therefore, overall cost is not reflected.
Conclusion
Findings reveal that treatment cost is higher in urban-located tertiary-level facilities. Given that a higher number of children in our sample were treated in urban-located hospitals, this highlights a substantial cost burden for the health system of the country. To address this issue, treatment services for children with severe pneumonia should be made more accessible in primary and secondary-level facilities. This approach could reduce overall treatment costs, as only complex cases would be referred to tertiary-level facilities. This would enable healthcare providers to manage more patients with the available resources, potentially improving equity in healthcare provision.
Data availability statement
Data are available upon reasonable request.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants. Deakin University Human Research Ethics Committee (DUHREC) (2018-149) approved the economic analysis protocol. The clinical trial (NCT02669654) protocol was approved by the Institutional Review Board of icddr,b (PR-14066). Written informed consent was gained from each respondent prior to data collection.
Acknowledgments
The International Centre for Diarrheal Disease Research of Bangladesh (icddr,b) is grateful to the Governments of Bangladesh, Canada, Sweden and the UK for providing core/unrestricted support. icddr,b acknowledges with gratitude the commitment of UBS Optimus Foundation and UNICEF, Switzerland, for funding the trial.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
X @marufa_sultana
Contributors MS, JJW and LG conceptualised the study. MS, NHA, MJC, ASGF, GJF and NG designed the study and data collection instruments, and supervised data collection. MS and NA supervised data collection and analysed data under supervision of JA-O, JJW and LG. MS drafted the initial manuscript and LG supervised each stage of drafting the initial manuscript. JJW, NHA, NA, ASGF, NA, GJF, NG, MJC, TA and LG critically reviewed and revised the manuscript. All authors approved the final manuscript. MS is guarantor.
Funding This work was jointly supported by UNICEF, Botnar Foundation, UBS Optimus Foundation and EAGLE Foundation, Switzerland (grant numbers: GR-01083; GR-01269). MS was supported by Deakin University Postgraduate Research (DUPR) scholarship.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.