Household Costs Associated with Hospitalization of Children with Severe Pneumonia in Quito, Ecuador

Abstract.

Pneumonia remains a leading cause of morbidity and mortality in young children. The total cost of pneumonia-related hospitalization, including household-level cost, is poorly understood. To better understand this burden in an urban setting in South America, we incorporated a cost study into a trial assessing zinc supplements in treatment of severe pneumonia among children aged 2–59 months at a public hospital in Quito, Ecuador, which provides such treatment at no charge. Data were collected from children’s caregivers at hospitalization and discharge on out-of-pocket payments for medical and nonmedical items, and on employment and lost work time. Analyses encompassed three categories: direct medical costs, direct nonmedical costs, and indirect costs, which covered foregone wages (from caregivers’ self-reported lost earnings) and opportunity cost of caregivers’ lost time (based on the unskilled labor wage in Ecuador). Caregivers of 153 children completed all questionnaires. Overall, 57% of children were aged less than 12 months, and 46% were female. Just over 50% of mothers and fathers had completed middle school. Most reported direct costs, which averaged $33. Most also reported indirect costs, the mean of which was $74. Fifty-seven reported lost earnings (mean = $79); 29 reported lost time (estimated mean cost = $37). Stratified analyses revealed similar costs for children < 12 months and ≥ 12 months, with variations for specific items. Costs for hospital-based treatment of severe pneumonia in young children represent a major burden for households in low- to middle-income settings, even when such treatment is intended to be provided at no cost.

INTRODUCTION

In many low- and middle-income countries (LMIC), pneumonia is a leading cause of morbidity and mortality, with the highest incidence among children younger than 5 years.¹ Community-acquired pneumonia (CAP) is the leading infectious cause of disease in this population,² responsible for 808,694 child deaths in 2017,³ approximately 15% of all child deaths globally.^2,3 Although pneumonia incidence is decreasing globally, it continues to be a leading cause of death in low-resource settings, where the vast majority of childhood pneumonia cases occurs (an estimated 138 million cases in 2015).⁴ In Latin America and the Caribbean, 80,000 children die annually from lower respiratory tract infections, over 85% of which are attributable to pneumonia and influenza-like diseases.⁵ In Ecuador, 12% of yearly deaths in children younger than 5 years are caused by pneumonia.^6,7 Although studies have shown that outpatient community-based treatment is effective and safe, and the WHO has modified its guidelines for treatment of severe pneumonia accordingly,⁸ hospitalizations of young children for treatment of severe pneumonia remain common globally because of potential complications and the need for oxygen therapy.⁹

The full economic cost of hospitalization for pneumonia treatment, encompassing costs both at the health system (e.g., providers) and household levels, is poorly understood. The cost burden for pneumonia diagnostics and treatment has been estimated to be $109 million per year globally;³ other studies have estimated the costs of treatment of pneumonia in different settings with varying results,^10–27 although these typically do not incorporate the economic toll on households. This burden imposes what may amount to substantial payments for consultations, hospital stay, tests, and medicines. In addition to incurring direct medical costs, households may bear substantial nonmedical costs and indirect costs, including transportation, lodging, food, and lost income in the form of lost wages and lost caregiver time, from a single pneumonia episode.^28,29 The few studies that have reported on the household-level cost of pneumonia in young children in LMIC have concluded that it can be high, particularly relative to household income, and represents a substantial economic burden. One 2008 study in northern Pakistan estimated a mean household cost of $18 for inpatient treatment of severe pneumonia, with medicines representing the greatest single cost.²⁹ Other studies found mean household cost estimates of US$12.54–27.28 in Kenya,³⁰ $41.35 in India,³¹ and $31.3 in Haripur, Pakistan, where costs varied depending on the type and level of facility.²⁸ These estimates highlight the substantial financial burden of severe pneumonia to households that often struggle to meet the most basic needs of family members.

Zinc, which is essential for healthy growth, development, and immune function, is commonly lacking in the diets of populations living in low-income settings.^32,33 Studies have shown that zinc can reduce both the frequency and severity of acute lower respiratory infections (such as pneumonia) and diarrhea.^34,35 To test the hypothesis that adjunctive zinc supplementation for pneumonia would reduce both duration of time to resolution of pneumonia symptoms and treatment failure, a team of researchers at Boston University and the Central University of Ecuador (Universidad Central del Ecuador) conducted the EcuaPAZ study, a double-blind, placebo-controlled, randomized trial, in Quito, Ecuador. A total of 450 children aged 2–59 months hospitalized for severe pneumonia at the study site³⁶ were given standard antibiotics and randomized to receive zinc supplements or placebo. Analysis of results showed no difference between groups in time to resolution of symptoms (the primary outcome) or treatment failure (the secondary outcome) (see Sempértegui et al.³² for full details).

Given the evidence from other LMIC of the substantial cost burden imposed on households from episodes of severe pneumonia in young children, and the dearth of similar evidence from Latin America, we undertook a cost study associated with the EcuaPAZ study. Although public hospitals in Ecuador provide such medical treatment at no charge, in practice, stock-outs of medical items and/or equipment failures and shortages result in patients having to obtain items at a personal cost at another hospital, clinic, or shop. In addition, households must cover a range of nonmedical costs and/or forego work-related income while their child is hospitalized. The objective of the cost study was to estimate household-level costs for hospitalization for treatment of severe pneumonia in young children at a large, urban public hospital setting in Quito, where medical treatment is intended to be provided for free to relatively poor families, similar to other economically vulnerable populations in Latin America. Both direct and indirect medical and nonmedical costs were included to contribute to the literature on the full economic burden of treatment in this population. To our knowledge, no study has yet investigated the household cost of treating severe pneumonia among young children beyond Kenya and South Asia.

MATERIALS AND METHODS

Study summary.

The primary study was a randomized, double-blind, placebo-controlled clinical trial conducted from February 2008 to April 2010 in Quito, Ecuador, at the Baca Ortiz Children’s Hospital.³² The Baca Ortiz Hospital is the main pediatric referral hospital in Ecuador, where the most common reason for hospitalization is acute lower respiratory infection. The cost study was added to the trial after enrollment had begun, with the primary aim of collecting household-level cost data from caregivers to better understand the economic burden of inpatient treatment of severe pneumonia in young children.

Children aged 2–59 months admitted to the Baca Ortiz Children’s Hospital with severe pneumonia were eligible for trial participation. Severe pneumonia was defined based on a modification of the WHO definition as follows: the presence of cough and/or chest wall indrawing, tachypnea (using age-dependent thresholds), hypoxemia (oxygen saturation < 90%), and at least one of the following signs of infection: rales, diminished breath sounds, bronchial breath sounds, or pleural rub.⁸ Exclusion criteria included parental refusal of informed consent, inability to take fluids, prior treatment with antibiotics for > 24 hours before admission, or any of the following diagnoses: marasmus, kwashiorkor, measles, pneumonia resulting from reported aspiration, hepatic or renal disease, neurologic disorders, acute diarrhea with dehydration, other viral infections, sepsis, congenital abnormalities, complicated pneumonia (e.g., lung abscess, pleural effusion, empyema, pneumatocele, and atelectasis), and severe anemia (hemoglobin < 8 g/dL). Caregivers provided written informed consent for a child’s participation.

Any caregiver of a child enrolled in the trial was eligible for the cost study. Once the latter was approved in early 2009, caregivers of each child in the trial were asked about interest in participating in the cost study. We used a nonprobability sampling approach, with caregivers selected conveniently by virtue of their relationship with children enrolled in the trial. We aimed to recruit one caregiver of each child who was enrolled in the trial until the conclusion of the trial (estimated to be roughly 1 year later), with a goal of recruiting a minimum of 100 caregivers, a number we deemed sufficient to generate robust findings based on other similar studies.^29,31

The study was approved by the Boston University Medical Center Institutional Review Board and the Ethics Committee of the Corporación Ecuatoriana de Biotecnología. All caregivers provided written informed consent before being enrolled into the study.

Data collection.

At enrollment into the primary study, trained study physicians collected information about each child’s basic demographics, current illness, and medical history from a caregiver. At the time of their child’s discharge from the Baca Ortiz Hospital, caregivers participating in the cost component of this study were also administered a structured verbal questionnaire.

Measures.

Direct medical questions queried out-of-pocket payments for medicine, diagnostic examinations, medical equipment, treatment procedures, bed charges, and all other costs associated directly with the hospitalization (“Did you pay out of pocket for medicine?” “How much did you pay total for the medicine?” “Did you pay for any diagnostic exams?” “Did you pay for any medical equipment?”). Direct nonmedical questions addressed transportation, including mode of transportation, minutes spent, and costs incurred (“What type of transportation did you use to bring your child to the hospital?” “How much did you pay in total for transportation while your child was in the hospital?”); communication with others while the child was hospitalized (“How did you communicate with your family that is not at the hospital with you?” “How much did you pay in total for communication?”); and costs for child care (“Who is taking care of your family in your absence?” “How much did you pay in total to the person that is taking care of your family?”). The questionnaire was pretested and modified to improve clarity of the questions. The funder was not involved in the development of the data collection instrument.

Indirect costs asked about the work of caregivers (“What type of work do you do?” “What type of work does the father of your child do?”), work maintenance during the child’s hospitalization (“Were you able to maintain your job during this time?” “Was the father able to maintain his job during this time?”), job loss (“Did anyone in the family lose their job because of the time they had to spend in the hospital? ”), income loss (“Can you estimate the amount of money you lost during these days by not working?” “Can you estimate the amount of money the father lost by not working?”), and other children (“Where are your other children?” “Did you have to take any of your children out of school while you were in the hospital?”).

Cost analysis.

The cost analysis focused on three main categories of costs for households associated with hospitalization of a child due to severe pneumonia (see Box 1): 1) direct medical costs, 2) direct nonmedical costs, and 3) indirect costs. The latter category is recommended by the WHO³⁷ and represented an effort to conduct a more comprehensive analysis of the economic burden borne by households than is obtained when considering direct payments alone. Ecuador uses U.S. dollars as its currency, so there was no need to convert currency. Any missing values on various questions are noted in the methods and results in the following paragraphs.

Box 1. Items included in estimating direct and indirect costs incurred by households because of hospitalization of a child with severe pneumonia in Quito, Ecuador.

A. Medical costs

 Diagnostics

 Medications

 Consultations

 Bed charges/hospital admissions

 Supplies

B. Nonmedical costs

 Transportation

 Food and lodging

 Communication

 Care of other children

 Miscellaneous household tasks

C. Indirect costs

 Lost earnings

 Opportunity cost of time

Direct costs were determined from self-reports of caregivers at the follow-up survey as described earlier. These encompassed payments made for diagnostic tests, medicines, inpatient consultations, and supplies related to a child’s pneumonia hospitalization. Direct nonmedical costs encompassed costs related to transportation, meals and accommodation for caregivers and other family members during the hospitalization, communication, the costs of childcare for children who remained at home during the sick child’s hospitalization, and miscellaneous costs (Box 1). Given a high number of missing values and “0” for some individual items, we calculated total costs for each category, as well as means and medians in two ways: a) means/medians excluding only missing values (but including “0” values) and b) means/medians excluding both missing values and “0” values (i.e., the denominator included only those participants who reported at least one cost > $0). The results of the two analyses produced similar results (hence the former is reported here); however, where the difference in means was > 10%, details on the results of the latter approach are provided in table footnotes. In addition, we included the values of small subsidies provided by the study team to caregivers who requested help with selected cost items (transportation, food, and communication costs).

Indirect costs took into account both foregone wages and the opportunity cost of caregivers’ lost time while the sick child was hospitalized. Lost wages were estimated on the basis of caregivers’ self-reported lost earnings that resulted directly from the hospitalization, including the time to transport the child from home to the hospital. The opportunity cost of lost time, representing the value of the caregivers’ time they otherwise would have spent engaging in normal daily unpaid activities, was estimated by (t × V), where t is the amount of time lost and V is the value of lost time; t was based on the time (in hours) reported by caregivers, which was converted to working days, and V is the unskilled labor wage in Ecuador (US$1.90 per hour). This wage was calculated from the International Labour Organization’s average monthly earnings of employees in Ecuador ($323 per month) in 2010 when the study was conducted.³⁸ These values were used to estimate the hourly wage lost due to hospitalization. We calculated totals and means/medians as described previously. Given the variability in wages across Ecuador, we conducted additional sensitivity analyses, using this cost with minimum and maximum ranges of ±50%, an approach that is often used when estimates are uncertain.³⁷

Household burden based on age of child.

Because of the higher prevalence of severe pneumonia among infants, we hypothesized that costs might be different in this group compared with older children. Therefore, we sought to determine whether households with younger children (aged less than 12 months) bore a higher burden than households with older children (aged 12 months or older). For this analysis, we disaggregated the data according to these age parameters and analyzed the total and mean costs by category as described above for the full sample. Within the two age-groups, we also used Pearson’s chi-squared test to explore differences in socioeconomic variables (e.g. parents’ education, crowdedness of living conditions) that might have had an effect on the cost data or severity of or time to resolution of symptoms. We also compared the average length of stay in the hospital between age groups with independent sample t-tests (significance threshold of P = 0.05).

Data entry and software.

Data were entered into an Excel database in Quito and analyzed in SPSS version 15.0 (SPSS, Inc., Chicago, IL) and SAS 9.4 (SAS Institute, Cary, NC).

RESULTS

Participants.

In the EcuaPAZ trial, 2,768 children were screened, of which 662 children met the eligibility criteria and 450 children were enrolled (225 children assigned to each study group—zinc and placebo) from February 2008 through April 2010. For the cost study, a total of 175 caregivers of children who were trial participants were approached during the enrollment period, and all agreed to participate. Among them, 153 completed the questionnaire before departing from the hospital, and, therefore, were included in the cost component of the study. Of those, 56.9% (87) had children who were younger than 12 months; 46.4% (71) of children were females and 53.5% (82) were males (see Table 1). The mean household size was approximately five family members. The sample population had low levels of education; although most mothers (53.6%) had completed middle school, very few had completed high school or attended some college (5.3%), or had a college degree or higher (4.0%). This distribution was similar for fathers; 34.3% (47) had completed elementary school, 57.7% (79) had completed middle school, 5.1% (7) had completed high school or some college, and 2.9% (4) had a college degree or higher. Households generally had one to two household members per room or > 2 per room (54.6% and 38.8%, respectively) (Table 1). Households with younger children tended to be more crowded (P = 0.031), and older children tended to be more underweight (P = 0.010).

Table 1.

Background characteristics of participants

	All children, N = 153		Children younger than 12 months, N = 87		Children ≥ 12 months, N = 66
	N	Mean/proportion	N	Mean/proportion	N	Mean/proportion	P-value
Characteristics of child’s family
Household size (members)*	152	5.0	86	5.2	66	4.6	0.03
Maternal education level†
Elementary school (0–6 years) (%)	56	37.1	29	33.7	27	41.5	0.58
Middle school (7–12 years) (%)	81	53.6	50	58.1	31	47.7
High school or some college (13–15 years) (%)	8	5.3	3	3.5	5	7.7
College degree or higher (16+ years) (%)	6	4.0	4	4.7	2	3.1
Paternal education level‡
Elementary school (0–6 years) (%)	47	34.3	27	35.5	20	32.8	0.52
Middle school (7–12 years) (%)	79	57.7	45	59.2	34	55.7
High school or some college (13–15 years) (%)	7	5.1	2	2.6	5	8.2
College degree or higher (16+ years) (%)	4	2.9	2	2.6	2	3.3
Crowdedness (home conditions, members in a household/bedrooms)*
< 1 per room (%)	10	6.6	4	4.7	6	9.1	0.03
1–2 per room (%)	83	54.6	41	47.7	42	63.6
> 2 per room (%)	59	38.8	41	47.7	18	27.3
Characteristics of the child
Age (months)	153	13.7	87	6.0	66	24.0	< 0.001
Age (months)
Female	71	14.3	40	5.6	31	25.5	< 0.001
Male	82	13.3	47	6.5	35	22.5
Height (%)§
Normal	119	82.1	69	81.2	50	83.3	0.17
Stunted‖	26	17.9	16	18.8	10	16.7
Weight (%)
Normal	122	81.3	75	88.2	47	72.3	0.04
Underweight¶	28	18.7	10	11.8	18	27.7
Characteristics of hospitalization (days)
Average length of stay	153	3.9	88	4.4	65	3.2	0.02

* 1 missing value.

† 2 missing values.

‡ 16 missing values.

§ 8 missing values.

‖ Stunted is defined as height-for-age more than two SDs below the WHO Child Growth Standards median.

¶ Underweight is defined as weight-for-age more then two SDs below the WHO Child Growth Standards median.

Direct and indirect costs of hospitalization for severe pneumonia.

The results for cost variables for all participants are displayed in Table 2. All participants (n = 153) reported one or more direct costs > $0. Most reported direct medical costs (n = 96, 62.3%), of whom the majority described treatment costs; over one-third of all participants (n = 57, 37.0%) recounted costs related to medication. Few declared costs for diagnostics or bed/hospital admission. All participants reported one or more direct nonmedical costs, of whom most (n = 142, 93.4%) described communication expenses; high proportions also reported transportation expenses (n = 123, 80.3%) and food and lodging (n = 86, 56.2%); few noted costs related to household tasks (n = 11, 7.1%) and care of other children besides the hospitalized child (n = 8, 5.2%). Nearly half of the participants reported indirect costs (n = 75, 49.0%), among whom most described lost earnings (n = 57, 76.0%); few reported indirect costs in terms of lost time (n = 29, 38.7%).

Table 2.

Direct and indirect costs (US$) for households with child hospitalized with severe pneumonia

	N*	Mean/proportion	Median†	SD	Minimum	Maximum
I. Direct costs
A. Medical
Diagnostics	20	35.9	16.0	41.0	2.5	154.0
Medications	57	14.5	11.0	12.9	2.0	50.0
Treatment‡	89	8.7	7.2	8.9	0.0	71.0
Bed charges/hospital admissions	4	10.4	9.25	8.5	3.0	20.0
Mean of direct medical costs§	96	23.5	–	–	–	–
B. Nonmedical
Transportation‖	123	4.8	1.0	11.1	0.0	80.0
Food and lodging¶	86	12.0	10.0	5.8	2.0	40.0
Communication#	142	3.0	1.8	2.9	0.1	15.0
Care of other children**	8	33.1	25.0	29.0	0.0	100.0
Miscellaneous household tasks	11	44.5	33.0	38.0	10.0	150.0
Mean of direct nonmedical costs††	153	18.3	12.3	24.4	0.0	165.0
Mean: direct costs‡‡	153	33.0	–	–	–	–
II. Indirect costs
Lost earnings	57	79.0	45.0	77.2	3.0	300.0
Opportunity cost of lost time§§	29	36.8	28.5	40.5	3.8	171.0
Opportunity cost of time, +50%‖‖	–	55.2	42.8	–	–	–
Opportunity cost of time, −50%‖‖	–	18.4	14.3	–	–	–
Mean: indirect costs¶¶	75	74.3	–	–	–	–
Mean of total costs	153	49.5	44.7	39.1	0.0	239.9

* N is the number of households of the total sample of 153 that reported a specific cost item of ≥ $0, unless otherwise explained (excludes missing values only). For any categories in which the mean cost between households that reported a cost > $0 and households reporting a cost ≥ $0 was greater than 10%, the values for the former are provided in footnotes.

† The median is also reported because of SDs that are higher than the mean.

‡ Among households that reported > $0 costs for treatment (n = 86), mean = 9.0, median = 7.4, SD = 8.8, minimum = .11, and maximum = 71.0.

§ Participants who reported at least one direct medical cost are included in this calculation. A total of n = 57 participants reported no direct medical costs.

‖ Includes one child older than 12 months who was given a $21 subsidy to cover transportation to another hospital.

¶ Includes 70 households that were given a subsidy as part of the study to cover food/lodging; 11 households were given $20 and 59 were given $10.

# Includes 58 households that were given a communication subsidy of $1.80.

** Among households that reported >$0 costs for care of other children (n = 7), mean = 37.0, median = 30.0, SD = 28.0, minimum = 10.0, and maximum = 100.0.

†† Participants who reported at least one direct nonmedical cost, including those given subsidies, are included in this calculation (N = 153).

‡‡ Participants who reported at least one direct medical or one direct nonmedical cost are included in this calculation (N = 153).

§§ The opportunity cost of lost time, representing the value of caregivers’ time that otherwise would have been spent engaging in normal daily unpaid activities, was estimated by t × V, where t represents the amount of time lost and V represents the value of lost time. t was based on the time (in hours) reported by caregivers, which was converted to working days (8 hours/day); V was estimated using the unskilled labor wage in Ecuador (US$1.90/hour); this value was calculated from the International Labour Organization’s average monthly earnings of employees in Ecuador ($323 per month) in 2010 when the study was conducted.

‖‖ Given the variability in wages across Ecuador, we conducted additional sensitivity analyses to estimate the opportunity cost of lost time, given minimum and maximum ranges of ± 50% for V.

¶¶ Participants who reported any lost earnings or lost time were included in these calculations. A total of n = 78 participants reported no such indirect cost.

The overall household mean cost associated with pneumonia hospitalization was $49.5. Of these, average direct costs came to $33.0. This total encompassed mean medical and nonmedical costs of $23.5 and $18.3, respectively. Among direct medical costs, diagnostics was the single greatest burden, $35.9 on average (n = 20), with a range of $2.5–$154.0 for those who bore this cost. The mean cost of the item reported by most participants (n = 89), for treatment, was $8.7. Among direct nonmedical costs, the highest mean cost was for miscellaneous household tasks ($44.5), which was reported by 11 participants, and ranged from $10.0 to $150. The average cost of the single item paid by most participants, for communication (n = 142), was $3.0.

Indirect costs averaged $74.3 for those who reported one or more indirect costs (n = 75). The mean value of lost earnings was $79.0 and ranged between $3.0 and $300.0 for those participants who reported a positive cost. For the opportunity cost of lost time, or the value of the caregivers’ time otherwise spent on normal daily activities, the mean estimated cost was $36.8 and ranged from $3.8 to $171.0 for those who provided a value of lost time > 0. The sensitivity analysis for the opportunity cost of lost time yielded mean estimates of $18.4–$55.2.

Household burden based on age of child.

The overall mean household cost for children aged < 12 months and ≥ 12 months (see Tables 3 and 4) was similar, at $49.7 (n = 87) and $49.3 (n = 66), respectively, but varied for specific cost components. For instance, the mean of direct costs was $33.1 versus $31.4 for younger children versus older children. However, direct medical costs were substantially higher among younger than older children (means = $27.6 and $16.1, respectively). This difference was primarily because of the relatively high cost of diagnostics (mean = $42.7) for the younger age-group, among whom 15/87 (17.2%) had these expenses during hospitalization. Conversely, direct nonmedical costs were higher for caregivers of older children; this group reported a mean of $24.7 in nonmedical expenses compared with $13.5 for caregivers of younger children. The specific items within the nonmedical cost category were all higher for households of older children (except for food and lodging, which was similar), although some particularly high costs, such as payments for household tasks (mean of $76.7 among caregivers of older children), were reported by very small numbers of caregivers (n = 3). For both groups, transportation was the most frequently cited cost (roughly 80% in both groups), averaging $3.7 for families of younger children compared with $6.2 for those of older children. Indirect costs followed a similar pattern: they averaged $91.5 for families of older children versus $61.5 for those of the younger group. The difference was most pronounced for lost earnings, for which the mean was $95.4 (reported by 27/66) among families of older children, ranging up to a maximum of $300 versus $64.3 (30/87) for families of younger children, with a maximum reported of $240.0.

Table 3.

Direct and indirect costs (US$) for households with children younger than 12 months hospitalized with severe pneumonia (N = 87)

	N*	Mean/proportion	Median†	SD	Minimum	Maximum
I. Direct costs
A. Medical
Diagnostics	15	42.7	14.0	45.3	6.0	154.0
Medications	37	14.8	11.0	13.9	1.6	70.5
Treatments	55	10.0	8.9	10.4	0.0	71.0
Bed charges/hospital admissions	2	3.3	3.3	0.4	3.0	3.5
Mean of direct medical costs‡	62	27.6	14.8	35.7	2.0	166.9
B. Nonmedical
Transportation	70	3.7	1.0	8.1	0.0	52.0
Food and lodging§	30	12.6	10.0	5.1	1.0	20.0
Communication‖	79	2.5	1.8	2.5	0.1	15.0
Care of other children	3	20.0	20.0	8.2	10.0	30.0
Miscellaneous household tasks	8	32.3	29.3	21.8	10.0	80.0
Mean of direct nonmedical costs¶	87	13.5	8.0	16.5	0	93.6
Mean: direct costs#	87	33.1	23.8	37.7	1.0	204.5
II. Indirect costs
Lost earnings	30	64.3	41.3	61.3	10.0	240.0
Opportunity cost of lost time**	21	34.4	30.4	38.2	3.8	172.9
Opportunity cost of time, +50%††	–	51.6	45.6	57.3	–	–
Opportunity cost of time, −50%††	–	17.2	15.2	19.1	–	–
Mean: indirect costs‡‡	43	61.5	–	–	–	–
Mean of total costs	87	49.7	44.5	39.8	0	239.9

* N is the number of households of the total sample of 87 that reported a specific cost item of ≥ $0, unless otherwise explained (excludes missing values only). For any categories in which the mean cost between households that reported a cost > $0 and households reporting a cost ≥ $0 was greater than 10%, the values for the former are provided in footnotes.

† The median is also reported because of SDs that are higher than the mean.

‡ Participants who reported at least one direct medical cost are included in this calculation. A total of n = 19 participants reported no direct medical costs.

§ Includes 23 children younger than 12 months who were given a $10 subsidy and four children younger than 12 months who were given a $20 subsidy to cover food/accommodation.

‖ Includes 29 children younger than 12 months who were given a $1.80 subsidy for communication.

¶ Participants who reported at least one direct nonmedical cost are included in this calculation. A total of n = 8 participants reported no indirect medical costs.

# Participants who reported at least one direct medical or one direct nonmedical cost are included in this calculation. A total of n = 3 participants reported no direct costs.

** The opportunity cost of lost time, representing the value of caregivers’ time that otherwise would have been spent engaging in normal daily unpaid activities, was estimated by t × V, where t represents the amount of time lost and V represents the value of lost time. t was based on the time (in hours) reported by caregivers, which was converted to working days (8 hours/day); V was estimated using the unskilled labor wage in Ecuador (US$1.90/hour); this value was calculated from the International Labour Organization’s average monthly earnings of employees in Ecuador ($323 per month) in 2010 when the study was conducted.

†† Given the variability in wages across Ecuador, we conducted additional sensitivity analyses to estimate the opportunity cost of lost time, given minimum and maximum ranges of ± 50% for V.

‡‡ Participants who reported any lost earnings or lost time were included in this calculation. A total of n = 44 participants reported no such indirect cost.

Table 4.

Direct and indirect costs (US$) for households with children aged ≥ 12 months hospitalized with severe pneumonia (N = 66)

	N*	Mean/proportion	Median†	SD	Minimum	Maximum
I. Direct costs
A. Medical
Diagnostics	5	15.3	18.0	10.6	2.5	30.0
Medications	20	13.7	10.5	11.1	1.0	42.0
Treatments	34	6.6	5.2	5.1	0.0	20.0
Bed charges/hospital admissions	2	17.5	17.5	3.5	15.0	20.0
Mean of direct medical costs‡	34	16.1	13.0	14.9	0.11	71.1
B. Nonmedical
Transportation§	53	6.2	1.0	13.9	0.0	80.0
Food and lodging‖	56	11.6	10.0	6.2	2.0	40.0
Communication¶	63	3.7	2.8	3.3	0.2	15.0
Care of other children#	4	46.3	42.5	36.0	0.00	100.0
Miscellaneous household tasks	3	76.7	50.0	52.5	30.0	150.0
Mean of direct nonmedical costs**	66	24.7	13.5	30.9	2.1	165.0
Mean: direct costs††	66	31.4	–	–	–	–
II. Indirect cost††
Lost earnings	27	95.4	45.0	90.0	3.0	300.0
Opportunity cost of time‡	8	44.4	24.7	44.1	5.7	136.8
Opportunity cost of time, +50%§§	–	66.6	37.1	66.1	–	–
Opportunity cost of time, −50%§§	–	22.2	12.4	22.0	–	–
Mean: indirect costs‖‖	32	91.5	–	–	–	–
Mean of total costs	66	49.3	45.6	38.5	2.1	177.4

* N is the number responding to the specific cost item with cost > $0 (excludes missing values only). N is the number of households of the total sample of 66 that reported a specific cost item of ≥ $0, unless otherwise explained (excludes missing values only). For any categories in which the mean cost between households that reported a cost > $0 and households reporting a cost ≥ $0 was greater than 10%, the values for the former are provided in footnotes.

† The median is also reported because of SDs that are higher than the mean.

‡ Participants who reported at least one direct medical cost are included in this calculation. A total of n = 25 participants reported no direct medical costs.

§ Includes one child older than 12 months who was given a $21 subsidy to cover transportation to another hospital.

‖ Includes 40 children older than 12 months who were given a $10 subsidy and seven children older than 12 months who were given a $20 subsidy to cover food/accommodation.

¶ Includes 29 children younger than 12 months who were given a $1.80 subsidy for communication.

# Among households that reported > $0 costs for care of other children (n = 3), mean = 61.2, median = 50.0, SD = 27.8, minimum = 35.0, and maximum = 100.0.

** Participants who reported at least one direct nonmedical cost are included in this calculation. A total of n = 7 participants reported no indirect medical costs.

†† Participants who reported at least one direct medical or one direct nonmedical cost are included in this calculation. A total of n = 4 participants reported no direct costs.

‡‡ The opportunity cost of lost time, representing the value of caregivers’ time that otherwise would have been spent engaging in normal daily unpaid activities, was estimated by t × V, where t represents the amount of time lost and V represents the value of lost time. t was based on the time (in minutes) reported by caregivers, which was converted to working days (8 hours/day); V was estimated using the unskilled labor wage in Ecuador (US$1.90/hour); this value was calculated from the International Labour Organization’s average monthly earnings of employees in Ecuador ($323 per month) in 2010 when the study was conducted.

§§ Given the variability in wages across Ecuador, we conducted additional sensitivity analyses to estimate the opportunity cost of lost time, given minimum and maximum ranges of ± 50% for V.

‖‖ Participants who reported any lost earnings or lost time were included in these calculations. A total of n = 34 participants reported no such indirect costs.

DISCUSSION

To our knowledge, this analysis provides important insights into the cost burden experienced by households for treatment of severe pneumonia among young children in South America, even when treatment is intended to be provided to relatively low-income families at no cost. Understanding this burden is critical because, although pneumonia incidence is decreasing globally,² it continues to be a leading cause of death in LMIC, including in Ecuador, where it accounts for 12% of yearly deaths in children younger than 5 years.^6,7 Recent research has estimated the economic burden of pneumonia on communities and nations as a whole,^10–14,19 but only a few studies have illuminated the burden of severe pneumonia in young children at the level of those individuals living closest to the child: households. The present study adds to the growing evidence that this burden is considerable, even when pneumonia treatment is intended to be provided at no charge. Several specific findings emerged that are noteworthy.

First, we found that total direct costs, including medical costs (e.g., diagnostics, medications, and treatments), were substantial. In Ecuador, the per capita health expenditure in 2007 was approximately $200, of which $78 was public spending.³⁹ The remaining amount ($122) was covered by out-of-pocket payments by individuals and private prepaid plans.³⁹ The $33.0 in average direct cost per hospitalization for child pneumonia that we estimated suggests that for those households who do experience this hospitalization, more than one-fourth of typical annual per capita expenditure is spent, on average, on out-of-pocket payments for a single episode of pneumonia-related hospitalization of a young child. For those who report a high indirect cost, the burden is even more substantial. This finding is consistent with other studies assessing the cost of pneumonia hospitalization, which found that a substantial portion of per capita expenditure is spent on treating pneumonia in outpatient settings and in hospitalizations.^28,30,31,40 An obvious question is: Why are families bearing these costs when treatment is intended to be free? In our team’s experience, hospitals often run out of medications (e.g., reagents and treatments), and functioning diagnostic tools may be unavailable; in acute situations, these items must be paid for at an outside laboratory or pharmacy. This highlights the need for more reliable supply of all aspects of treatment at hospitals, including those providing free care to low-income patients.

Beyond medical costs, we found that families spent an average of $18.3 on nonmedical items during hospitalizations. These encompassed transportation, lodging near the hospital while their child was being treated, meals, child care for other children in the family, and various household tasks while parents were away. Average payments for the latter two categories were particularly high: for those who reported these expenses, the average was $33.1 (n = 8) and $44.5 (n = 11), respectively, much higher than the mean of all direct costs (which included participants who reported any single cost). Although these amounts seem small compared with providers’ costs, they represent a considerable burden for many households in Ecuador, particularly those at lower income strata. These results align with those found in other settings in Africa and South Asia, where researchers have shown that poor households often cope with a substantial financial burden that, in large part, is due to indirect payments outside of hospitals and clinics.^28–31

In addition, we found that indirect costs were quite high, higher than direct costs for households of all children, irrespective of age. For those who reported one or more indirect cost (most participants), the mean cost was much higher than direct costs ($74.3 compared with $33.0). One of the striking results was the very high reported lost earnings ($79.0 on average, for those who reported this cost, and much higher for households of the older age-group) among caregivers of hospitalized children. Although we did not collect participants’ incomes, the average wage for unskilled labor that we used to estimate the opportunity cost of lost time ($323) suggests that indirect costs (including foregone wages) can easily approach one-fourth of the monthly wages for someone who may be the family breadwinner; this is on top of the roughly one-tenth spent on direct costs. Similarly, the opportunity cost of lost time represents a substantial and often hidden burden that falls on caregivers of hospitalized children. The few other published studies of household-level costs of child pneumonia echo this conclusion. One found that taken together, the economic burden of medical costs and indirect costs stemming from a hospitalization contributed to a 30-fold rise in overall household costs when compared with community-based management of severe pneumonia; the indirect costs alone accounted for 37% of the burden on households.²⁸ In Kenya, the economic burden on families exceeded an entire month of normal spending once direct and indirect costs were considered.³⁰

These substantial direct and indirect costs are especially concerning considering the relatively low socioeconomic status of many of the families whose children are hospitalized for treatment of severe pneumonia. Although we did not collect individual income data, most caregivers who participated in our study reported an education level of middle school or less, indicating that their incomes were likely relatively low. Previous studies have shown that pneumonia, like other infectious diseases, generally affects the poorest populations globally.⁴¹ Families living in poverty often have few financial and material resources to prevent the occurrence of severe pneumonia and to treat it in community-based settings when it does occur. Our study adds to the evidence showing that household-level costs associated with hospitalization of young children for pneumonia place an enormous burden on those households least able to cope, potentially exacerbating the cycle of poverty due to decreased productivity, lost wages, and reduced earning potential. As noted earlier, it is widely accepted that outpatient treatment is effective and safe, and the WHO has modified its guidelines for treatment of severe pneumonia accordingly,⁸ but hospitalization of young children for treatment of severe pneumonia remains common.

We also found that households with younger children paid significantly more for diagnostic tests and for all direct medical costs combined. This cost is consistent with other studies that have found that the treatment cost of infant pneumonia increases significantly when treatment is provided on an inpatient, compared with outpatient, basis.³¹ We also found the opposite pattern in burden for direct nonmedical costs and for indirect costs. This suggests that families with slightly older children may face a quite different financial burden when considering nonmedical costs, but because of small numbers for some specific items such as child care and household tasks, we cannot be confident of this trend and urge further research on this question.

We wish to highlight that although few studies have focused on the economic cost to households of severe childhood pneumonia, such studies (including ours) only consider costs that can be quantified in monetary terms. However, we know that childhood pneumonia creates not only financial but also psychological stress for families. Although data are sparse, some studies have tried to demonstrate the broader toll that child illness can place on a family, including the negative impacts of reduced quality of life on parents, an increase in stress and anxiety, a significant loss of work days, and the complicated process of determining where to seek care.^42,43 A study in Israel found that CAP treatment in children younger than 3 years resulted in loss of routine and decreased quality of life, with quality of life declining as treatment intensified from primary health clinic to pediatric emergency department to pediatric ward.⁴² We want to recognize that when a child falls ill with a life-threatening illness, the burden on parents can be overwhelming—going far beyond monetary costs.

We note several study limitations. First, costs were assessed by caregivers’ self-report, rather than from the hospital, and may have been influenced by recall bias. However, cost data were collected at hospital discharge, and hospitalization is a dramatic and costly event, so we believe that caregivers’ reports were reasonably reliable. Moreover, cost-related data collection was nested within a larger randomized control trial, which resulted in rigorous, high quality data of hospitalized children with confirmed pneumonia. We were able to collect detailed information regarding the hospitalization, employment, child care, transportation, and other costs, which improved the accuracy of the direct and indirect costs. Second, the study was conducted at one hospital only. Although residents from many parts of the greater Quito area come to the study hospital for treatment, we cannot generalize findings to the rest of Ecuador. Third, we did not collect income data from caregivers, so we were unable to estimate with a high degree of precision the opportunity cost of the time spent in the hospital. Hence, we underscore the uncertainty of some of our indirect cost estimates. Finally, the study definition differed slightly from the WHO definition that was in use at the time by including hypoxemia and at least one sign of infection on pulmonary examination. These are relatively minor alterations of the WHO definition, so we feel confident that our data on cost of severe pneumonia can be compared with previous studies carried out in South Asia.

CONCLUSION

The economic burden of hospitalization for childhood pneumonia in Ecuador is substantial, including considerable direct costs (e.g., medical expenses) and indirect costs (e.g., payments for transportation and accommodation). Apart from these financial expenses, households can also bear an extensive cost due to lost wages and the opportunity cost of time, which contributes to the overall burden on households of a hospitalization compared with community management of severe childhood pneumonia. Since the completion of the study in 2010, the Ministry of Public Health in Ecuador has implemented several strategies designed to reduce the burden of pediatric pneumonia. In 2010, a 10-valent pneumococcal conjugate vaccine was introduced. In 2014, national guidelines for the prevention of pneumonia in children were released,⁴⁴ and in 2017, the ministry released new guidelines for the treatment of pneumonia in children aged 3 months to 15 years. The guidelines emphasized avoiding the routine use of antibiotics in preschool children, because they often have viral infections, and the use of outpatient oral amoxicillin for treatment of non-severe CAP when a bacterial etiology is strongly suspected.⁴⁵ Such strategies have the potential to reduce this substantial economic burden of childhood pneumonia in Ecuador and other similar settings in South America.