Investiaión oriinal / Oriinal researh Cost-effetiveness of improvin pediatri hospital are in Niaraua Edward I. Brouhton, 1 Ivonne Gomez, 1 Osar Nuñez, 1 and Yudy Won 1 Suested itation Brouhton EI, Gomez I, Nuñez O, Won Y. Cost-effetiveness of improvin pediatri hospital are in Niaraua. Rev Panam Salud Publia. 2011;30(5):453 60. abstrat Key words Objetive. To determine the osts and ost-effetiveness of an intervention to improve quality of are for hildren with diarrhea or pneumonia in 14 hospitals in Niaraua, based on expenditure data and impat measures. Methods. Hospital lenth of stay (LOS) and deaths were abstrated from a random sample of 1 294 linial reords ompleted at seven of the 14 partiipatin hospitals before the intervention (2003) and 1 505 reords ompleted after two years of intervention implementation ( post-intervention ; 2006). Disability-adjusted life years (DALYs) were derived from outome data. Hospitalization osts were alulated based on hospital and Ministry of Health reords and private setor data. Intervention osts ame from projet aountin reords. Deision-tree analysis was used to alulate inremental ost-effetiveness. Results. Averae LOS dereased from 3.87 and 4.23 days pre-intervention to 3.55 and 3.94 days post-intervention for diarrhea (P = 0.078) and pneumonia (P = 0.055), respetively. Case fatalities dereased from 45/10 000 and 34/10 000 pre-intervention to 30/10 000 and 27/10 000 post-intervention for diarrhea (P = 0.062) and pneumonia (P = 0.37), respetively. Averae total hospitalization and antibioti osts for both dianoses were US$ 451 (95% redibility interval [CI]: US$ 419 US$ 482) pre-intervention and US$ 437 (95% CI: US$ 402 US$ 464) post-intervention. The intervention was ost-savin in terms of DALYs (95% CI: US$ 522 US$ 32 per DALY averted) and ost US$ 21 per hospital day averted (95% CI: US$ 45 US$ 204). Conlusions. After two years of intervention implementation, LOS and deaths for diarrhea dereased, alon with LOS for pneumonia, with no inrease in hospitalization osts. If these hanes were entirely attributable to the intervention, it would be ost-savin. Hospital osts; hospitals, pediatri; ost-benefit analysis; diarrhea; pneumonia; delivery of health are; Niaraua. 1 University Researh Co., LLC/United States Aeny for International Development, Health Care Improvement Projet, Bethesda, Maryland, United States of Ameria. Send orrespondene to: Edward I. Brouhton, ebrouht@jhsph.edu; ebrouhton@ur-hs.om Pneumonia and diarrhea are responsible for the hihest burden of morbidity and mortality amon hildren in Niaraua (1, 2). Sinifiant aps have been found between the linial manaement of hildren in referral hospitals with these two dianoses and that reommended in the World Health Oranization (WHO)/UNICEF Interated Manaement of Childhood Illness (IMCI) stratey for low-inome ountries in eneral (3) and Niaraua in partiular (4). To redue these aps, Niaraua s Ministry of Health (MINSA) worked with the U.S. Aeny for International Development (USAID), throuh its Quality Assurane Projet (QAP), and UNICEF, to desin and implement a hospital-based intervention based on the quality improvement (QI) ollaborative approah. 2 Known as the Pediatri Hospital Improvement (PHI) Collaborative, the intervention was launhed in 2004 at 14 reional hospitals and foused on hildren admitted to the pediatri ward with 2 Use of a shared learnin approah amon a lare number of quality improvement teams who work toether toward the same aim to rapidly ahieve sinifiant improvements in the proesses, quality, and effiieny of servies (5). Rev Panam Salud Publia 30(5), 2011 453
Oriinal researh Brouhton et al. Cost-effetiveness of improvin pediatri hospital are in Niaraua severe ases of diarrhea or pneumonia. Intervention objetives, identified durin a 6-month evaluation onduted at six hospitals in 2003, were desined to address various quality-of-are problems, inludin: Lak of a standardized dianosis and lassifiation system based on expliit linial riteria, standard definitions for identifyin patients with severe dehydration, and uniform uidelines for treatin those patients; Inappropriate presription of antibiotis and other mediations; No speified room for lose, onstant monitorin of ritially ill hildren; Poor reonition of daner sins by linial staff; Inadequate triae system for patients seen in the emereny room; Inadequate written uidelines suh as linial alorithms and treatment protools to diret medial and nursin staff to evidene-based are; Poor transportation system for transferrin ritially ill hildren to referral failities; and Poor ability to identify the nutritional status of patients. The QI stratey inluded eiht omponents: 1) introduin the WHO/UNICEF referral are manual (6) to standardize the linial approah to treatment of hildren with severe diarrhea or pneumonia at the hospital level; 2) definin quality indiators to measure ompliane with standards and identify performane aps; 3) formin QI teams to promote linial pratie hanes that would lead to improved quality of are; 4) establishin mehanisms for prioritizin the are of hildren who ome to emereny rooms; 5) hanin hospital oranization so ritially ill hildren are onentrated in a sinle hospital environment to allow loser and ontinuous monitorin; 6) developin skills in medial and nursin staff to identify, manae, and monitor for daner sins; 7) improvin adherene to standardized presribin praties for antibiotis; and 8) involvin loal authorities in supportin quality ativities. While routine monitorin of quality indiators by PHI Collaborative hospital teams showed that the intervention was assoiated with a deline in ase fatality rates for diarrhea and pneumonia as well as an inrease in ompliane with linial protools (inludin the appropriate use of antibiotis) and redued hospital stays, its osts and ost-effetiveness were not known. To fill this ap, the authors of the urrent study used data on the intervention s expenditures and various measures of its impat to determine its overall ost and eonomi effiieny. This information ould help Niarauan deision-makers determine if nationwide implementation of this type of intervention would be affordable and ost-effetive ompared to other strateies to improve are. MATERIALS AND METHODS The study was based on two years of intervention implementation (2004 2005). Pre-/post-intervention evaluations were onduted to determine intervention impat. The researh team examined diarrhea and pneumonia inpatient medial reords and admissions reisters ompleted at partiipatin hospitals 1) before the intervention (in 2003) ( preintervention ), and 2) after two years of its implementation (in 2006) ( postintervention ). Expenditure data were obtained from hospital and MINSA reords and private setor alulations (for hospitalization osts) and UNICEF and USAID/QAP aountin reords (for intervention osts). Misellaneous trainin osts (e.., transportation, food, and housin) were based on author estimates. Impat measures and expenditures were defined aordin to the riteria of MINSA and USAID/QAP. Samplin The 14 hospitals partiipatin in the PHI Collaborative did not bein the intervention simultaneously. Therefore, to ensure the onsisteny of the data, the study sample was limited to hospitals where QI teams had implemented improvement yles and monitored the indiators of interest for at least two years (i.e., sine the intervention s launh in 2004). The seven qualifyin hospitals were Chinandea, Madriz, Estelí, Boao, Granada, Masaya, and La Trinidad. The authors randomly seleted hospital reords from all patient harts available for review for both the pre- and post-intervention evaluations. For the pre-intervention evaluations, 647 linial reords were seleted for both diarrhea and pneumonia. For the post-intervention evaluations, 750 reords were seleted for diarrhea and 755 reords for pneumonia. A total of 2 799 reords were examined out of the 14 833 ase reords (9 938 diarrhea and 4 895 pneumonia) available at the seven hospitals inluded in the study. Samplin was stratified by hospital. A sample of 90 110 reords was taken from eah hospital. The samplin number was determined aordin to USAID/QAP monitorin requirements. Cases of severe diarrhea and pneumonia were identified based on IMCI definitions (7). Patients with a simultaneous dianosis of diarrhea and pneumonia were exluded from the sample. Cost data Total osts were alulated retrospetively based on data from 1) hospital reords, MINSA, and private setor alulations (for hospitalization osts) and 2) UNICEF and USAID/QAP aountin reords (for intervention osts). Hospitalization osts inluded the salaries of the dotors, nurses, and anillary staff required to staff the pediatri wards, alulated aordin to MINSA salary rates; hospital bed osts, inludin food, water, medial supplies, eletriity, and other utilities, alulated aordin to rates established in private hospitals; the amount paid for antibiotis presribed and onsumed durin hospitalization, alulated aordin to the MINSA prie list of basi mediines; and the amount paid for laboratory blood tests for all patients (pneumonia and diarrhea), a hest radioraph for eah pneumonia patient, and a feal laboratory test for eah diarrhea patient, aordin to MINSA reords. For items that are not aounted for in Niarauan publi hospitals (e.., linial line items), estimated osts were based on those established for the private setor (whih are onsidered a reasonable approximation to osts in the publi setor). All osts were reorded in Niarauan órdobas, onverted to U.S. dollars usin the January 2006 exhane rate (C$17 per US$1) (8) and adjusted to 2010 by applyin an annual inflation rate of 3%. Effetiveness measures The three measures of the QI intervention s impat on ases of diarrhea or pneumonia were 1) hospital lenth of stay (LOS), 2) disability-adjusted life years or DALYs (years of potential life lost due to premature mortality and years of produtive life lost due to dis- 454 Rev Panam Salud Publia 30(5), 2011
Brouhton et al. Cost-effetiveness of improvin pediatri hospital are in Niaraua Oriinal researh TABLE 1. Inputs and distributions of deision tree preditin ost-effetiveness of pediatri are quality improvement (QI) intervention at seven reional hospitals in Niaraua, 2004 2005 a Pre-intervention Post-intervention Estimate 95% CI b PD Estimate 95% CI PD Lenth of stay (days) d Diarrhea 3.87 3.62 4.12 Normal 3.55 3.31 3.80 Normal Pneumonia 4.23 4.04 4.43 Normal 3.94 3.74 4.16 Normal DALYs e averted per death averted Diarrhea 31.05 31.05 Pneumonia 31.05 31.05 DALYs averted per aute ase averted Diarrhea 0.011 0.011 Pneumonia 0.013 0.013 Cost per hospital bed-day f Diarrhea 93.30 84.28 102.32 Normal 93.30 84.28 102.32 Normal Pneumonia 99.12 89.52 108.72 Normal 99.12 89.52 108.72 Normal Cost of antibiotis f Diarrhea 8.80 7.94 9.66 Normal 11.80 10.64 12.96 Normal Pneumonia 13.70 12.36 15.04 Normal 14.00 12.63 15.37 Normal Case fatality ratio d Diarrhea 0.0045 0.0023 0.0078 Binomial 0.0030 0.0011 0.0061 Binomial Pneumonia 0.0034 0.0020 0.0054 Binomial 0.0027 0.0015 0.0045 Binomial Proportion of ases that are diarrhea d 0.33 0.32 0.34 Binomial Cost of QI intervention 45 124 Total number of patients over two years of intervention d 14 833 Cost of QI intervention per patient admitted 3.04 a All osts in U.S. dollars (based on January 2006 exhane rate (C$17 per US$1) (8) adjusted to 2010 usin 3% annual inflation rate). b CI: redibility interval. PD: probability distribution. d Data from patient reords, reisters, and MINSA. e DALYs: disability-adjusted life years; ae weihtin and a life expetany of 73 (10) were applied to DALYs for mortality alulations. f Data from hospital and private setor ost alulations. Data from UNICEF and U.S. Aeny for International Development (USAID) / Quality Assurane Projet (QAP) aountin reords. ability), and 3) deaths. Data on LOS and mortality were olleted from the patient reords sampled in the partiipatin hospitals. DALYs for aute illness were estimated usin disability weihts assined to severe ases of pneumonia and diarrhea, obtained from Murray and Lopez (9), and assumin averae ae of onset was 2 years. Ae weihtin, a disount rate of 3% per annum, and a life expetany of 73 (10) were applied to DALYs for mortality alulations aordin to a standard methodoloy that used similar eonomi analyses without reard to ender (11, 12) (Table 1). The total number of patients admitted for eah of the two dianoses was obtained from hospital admission reistries. Analysis FIGURE 1. Analyti framework for deision tree preditin inremental osteffetiveness of pediatri are quality improvement (QI) intervention at seven reional hospitals in Niaraua, 2004 2005 Pre-QI intervention (business as usual) Diarrhea Pneumonia Diarrhea Death Survival Death Survival Death Deision-tree analysis was used to ompare the QI intervention stratey (improved ompliane with linial uidelines) to business-as-usual (preintervention linial manaement of diarrhea and pneumonia ase) in a sinle iterative model (Fiure 1). The inremental ost-effetiveness ratio (ICER), Post-QI intervention Pneumonia Survival Death Survival Rev Panam Salud Publia 30(5), 2011 455
Oriinal researh Brouhton et al. Cost-effetiveness of improvin pediatri hospital are in Niaraua defined as the inremental hane in ost divided by the inremental hane in effetiveness, was alulated by omparin indiators at the seven partiipatin hospitals pre- and post-intervention. In terms of DALYs, the ICER represented the ost per additional DALY averted ($/DALY) assoiated with the QI stratey. The lower the ICER value, the better the relative ost-effetiveness of the are delivered post-intervention ompared to that provided pre-intervention. A positive ICER value indiates that an additional expenditure is required to ahieve better health outomes, whereas a neative ICER value means the intervention is assoiated with better health outomes and a derease in overall ost and is thus desinated as ost-savin. The equation used to alulate the ICER (inremental ost/averae number of inremental DALYs) is shown below. Cost of [diarrhea/pneumonia] ase after the intervention ost of ase before the intervention Averae number of DALYs after the intervention averae number of DALYs before the intervention TABLE 2. Itemized osts (trainin, materials, and tehnial assistane) of pediatri are quality improvement (QI) intervention at seven reional hospitals in Niaraua, 2004 2005 a Cost per Total % of session b No. of No. of ost total Item (US$) hospitals sessions (US$) osts Soure Trainin Transportation 47 7 6 1 974 4.4 Food and lodin Sessions 1 4 328.5 7 4 9 200 20.4 Sessions 5 6 47 7 2 660 1.5 Partiipant supplies (notebooks et.) 14 7 6 600 1.3 Hospital personnel (150 nurse-days + 150 physiian-days) 7 072 15.7 d Subtotal 19 506 43.2 Materials (printin and distribution) IMCI e trainin methodoloy uide 116.7 7 4 3 268 7.2 f, IMCI hospital proedures manual 7 287 16.1 f IMCI standards and indiators uide 2 248 5.0 IMCI pediatri emereny are manual 2 100 4.7 Subtotal 14 903 33.0 Tehnial assistane for implementation of operational hanes 10 715 23.7 Total 45 124 100.0 a All osts in 2010 U.S. dollars (based on January 2006 exhane rate (C$17 per US$1) (8) adjusted to 2010 usin 3% annual inflation rate). b 30 partiipants per session. Sum of author estimates of ost per individual. d Ministry of Health hospital personnel salaries list. e IMCI: Interated Manaement of Childhood Illness (stratey for pediatri are reommended by World Health Oranization and UNICEF). f UNICEF aountin reords. U.S. Aeny for International Development (USAID)/Quality Assurane Projet (QAP) aountin reords. Beause the outome data used in the urrent study were olleted from a random sample of ases from seven different hospitals, the samplin distribution assoiated with the ICER point estimate (PE) (the averae of multiple ICER point estimates, representin estimated hospitalization and antibioti osts, aross multiple data sets) needed to be taken into aount to determine a redibility interval (CI). The authors used Monte Carlo simulations in whih repeated samplin from the distributions of all input variables was used to alulate a probability distribution for the ICERs. This allowed for alulation of a 95% CI. One-way probabilisti sensitivity analysis was used to rank the variables by their relative importane to the osteffetiveness result (in terms of DALYs) by usin eah of the variables in turn and performin Monte Carlo simulations to determine CIs based on the deree of unertainty for eah speifi variable. Appropriate probability distributions had to be assined to the input variables as some were not statistially sinifiant at the alpha = 0.05 level. These distributions were determined based on the study data (Table 1). RESULTS Averae LOS dereased from 3.87 days pre-intervention to 3.55 days postintervention for diarrhea ases (P = 0.078) and from 4.23 days pre- intervention to 3.94 days post- intervention for pneumonia (P = 0.055). The ase fatality ratio dereased from 45 per 10 000 preintervention to 30 per 10 000 postintervention for diarrhea (P = 0.062) and from 34 per 10 000 pre-intervention to 27 per 10 000 post-intervention for pneumonia (P = 0.37) (Table 1). Intervention osts are listed in Table 2. Tehnial assistane, whih aounts for almost 25% of the total ost of the QI intervention, inluded all work done by USAID/QAP personnel to implement the eiht stratey omponents. It did not inlude the time ost of MINSA administrators or hospital offiials beause their roles in these ativities were onsidered part of their normal work duties and therefore did not inur additional ost to MINSA. The time ost of hospital personnel attendin the trainin sessions was inluded, even thouh this did not inur any additional diret ost to MINSA, beause attendane in the sessions took these health workers away from their reular duties, and whether or not relief staff were brouht in to ompensate for this absene it was important to aount for this potential loss in produtivity. Trainin sessions onsumed the reatest proportion of total expenditures (43.2%). Costs per day for hospitalization (exludin antibiotis) remained unhaned (pre- versus post-intervention) at $93.30 and $99.12 per day for diarrhea and pneumonia ases respetively. The ost of antibiotis per day inreased from $8.80 to $11.80 for diarrhea and from $13.70 to $14.00 for pneumonia. The averae total ost of hospitalization and antibioti mediation for both dianoses was $451 pre-intervention and $437 post-intervention. In terms of DALYs, the QI intervention was ost-savin ompared to business-as-usual (95% CI: $522 $32 per DALY averted; Table 3). When the variability of the result is plotted on a ost-effetiveness aeptability urve, the intervention is 100% ertain to be ost-effetive when the health are 456 Rev Panam Salud Publia 30(5), 2011
Brouhton et al. Cost-effetiveness of improvin pediatri hospital are in Niaraua Oriinal researh TABLE 3. Inremental ost-effetiveness of pediatri are quality improvement (QI) stratey versus business-as-usual at seven reional hospitals in Niaraua, 2004 2005 a Total Inremental ost ost Total DALYs 95% CI d Stratey (US$) (US$) DALYs b averted ICER (US$) Business-as-usual (pre-intervention ase manaement) 451 NA e 0.129 NA NA NA QI intervention (improved ompliane with linial standards) 437 14 0.069 0.060 Cost- $522 $32 savin a All osts in 2010 U.S. dollars (based on January 2006 exhane rate (C$17 per US$1) (8) adjusted to 2010 usin 3% annual inflation rate). b DALYs: disability-adjusted life years; ae weihtin and a life expetany of 73 (10) were applied to DALYs for mortality alulations. ICER: inremental ost-effetiveness ratio (additional ost per additional DALY averted). d CI: redibility interval. e NA: not appliable. FIGURE 2. Cost-effetiveness aeptibility urve for pediatri are quality improvement (QI) intervention at seven reional hospitals in Niaraua, 2004 2005 1,00 0,98 Probability 0,96 0,94 0,92 0,90 0 50 100 150 200 250 300 350 400 Willinness to pay (US$) TABLE 4. Probabilisti one-way sensitivity analysis of variables for deision tree preditin ost-effetiveness of pediatri are quality improvement (QI) intervention at seven reional hospitals in Niaraua, 2004 2005 Variable 95% CI a ($/DALY) b Lenth of hospital stay Pneumonia 420 30 Diarrhea 375 74 Case fatality ratio Pneumonia 284 185 Diarrhea 283 193 Cost of antibiotis Pneumonia 319 133 a CI: redibility interval. b Cost per additional disability-adjusted life year averted (in 2010 U.S. dollars, based on January 2006 exhane rate (C$17 per US$1) (8) adjusted to 2010 usin 3% annual inflation rate). Ae weihtin and a life expetany of 73 (10) were applied to DALYs for mortality alulations. funder is willin to pay $200 or more per DALY averted (Fiure 2). In terms of hospital days, the intervention ost $21 (per hospital day averted) (95% CI: $45 $204). One-way probabilisti sensitivity analysis is shown in Table 4. The larer the redibility interval, the reater the unertainty assoiated with the variable and therefore the reater its importane to the ost-effetiveness result. LOS variables had the reatest effet on the result, followed by ase fatality ratios and ost of antibiotis. Unertainty in the ost of hospitalization had a small impat on the result, while unertainty in the ost of the QI intervention and the ratio of diarrhea to pneumonia ases had almost no effet on the result. DISCUSSION This study found lose to statistially sinifiant dereases in LOS (for diarrhea and pneumonia) and in deaths (for diarrhea) that were assoiated with the QI intervention. Calulation of osteffetiveness in terms of DALYs showed that the intervention was ost-savin, indiatin a hih likelihood that the QI intervention improves health outomes and dereases osts to the hospital system. This result explains the derease in ase fatality rates for both diarrhea and pneumonia that was observed followin the QI intervention. In terms of LOS, the ICER was $21, indiatin that dereasin the averae LOS would most likely result in a small additional ost. The fat that the ICER was Rev Panam Salud Publia 30(5), 2011 457
Oriinal researh Brouhton et al. Cost-effetiveness of improvin pediatri hospital are in Niaraua positive in terms of LOS but neative (i.e., ost-savin) in terms of DALYs an be attributed to the sope of the DALYs metri (whih onsiders deaths, and aute illness) versus that of the LOS metri (whih treats all dishared patients as one and the same reardless of the outome). The LOS ICER is presented to show the intervention s relative ost-effetiveness in finanial terms reardless of whether or not it meets the riteria for DALY alulations. Even when onsiderin the point on the ost-effetiveness aeptability urve at whih there is 100% ertainty of osteffetiveness (a willinness to pay of $200 per DALY averted), the ICER is hihly favorable. WHO defines a health intervention as very ost-effetive if its ICER is less than the GDP per apita of the population it serves (13). As Niaraua s per apita GDP is $2 900, the PHI Collaborative intervention would be onsidered very ost-effetive (14). Comparin the effiieny of the QI stratey used in this intervention to strateies used in other health interventions in Niaraua or other Central Amerian ountries is diffiult iven the pauity of eonomi evaluations of health prorams from the reion (15). To the best of the authors knowlede, no other ost-effetiveness studies have been onduted to evaluate interventions for pediatri diarrhea or pneumonia ases. The PHI Collaborative intervention appears to be sinifiantly more ost-effetive than ervial aner sreenin in Niaraua, reported to ost $3 700 per DALY averted (16), and ompares favorably with rotavirus vaination for diarrheal disease prevention in both Honduras, reported to ost $269 per DALY averted (17), and Northeastern Brazil, based on another eonomi analysis that showed its effets on osts for diarrhea were not ost-savin (18). Diarrhea and pneumonia are the two most ommon auses of infant death worldwide, aountin for 40% of all mortality in hildren under 5 years old (19, 20), and remain sinifiant publi health problems in Niaraua, where the infant mortality rate is 27/1 000 (21). Any strateies that help address the hih burden of disease resultin from these two onditions are likely to have a sinifiant impat on the overall effetiveness of the health system. Money saved due to the sliht derease in averae LOS that ourred over the ourse of the intervention was partially offset by the inrease in expenditure for antibiotis that resulted from improved ompliane with linial uidelines. Before the intervention, presription praties for antibiotis in Niaraua were reported to be enerally poor, leadin to suboptimal outomes and an inrease in the risk of resistane (22). Suboptimal praties inluded the presription of heaper antibiotis that in many ases were not reommended. The intervention addressed this issue by standardizin presribin praties and failitatin the use of more appropriate (and often more expensive) antibiotis for diarrhea and pneumonia. Therefore, while this part of the intervention ahieved the desired result of improved ompliane with presribin standards, it also led to an inrease in the averae ost per patient for antibiotis. Sensitivity analysis showed that the un - ertainty in LOS input variables had the reatest impat on the ost-effetiveness result. Given that the differenes between pre- and post-intervention LOS were not statistially sinifiant, ahievin a more aurate estimate of these inputs by usin a larer sample size would have yielded a more preise result. By definition, ollaborative improvement is a temporary intervention desined to 1) address speifi problems in the quality of servies and 2) institutionalize the hanes made in are proesses to ensure sustainability of the results. One this ours, QI teams either move onto another set of problems or disband. If that definition holds true, favorable results would ontinue to be ahieved upon ompletion of the PHI Collaborative intervention at the tareted failities, without further intervention and expense. If this were the ase, the QI intervention stratey would prove more ost-effetive when onsidered over a loner time period. Limitations While no other interventions were implemented to address the quality of diarrhea and pneumonia are in Niarauan hospitals either before or durin the PHI Collaborative intervention, the lak of a ontrol roup in this study desin leaves open the possibility that the hanes deteted in the pre-/post-intervention omparisons ould have been due to fators other than the intervention. Further studies usin a ontrol roup or an analoous desin would help determine how muh improvement was attributable speifially to the PHI Collaborative initiative. Other than the start-date of their respetive intervention ativities, there were no substantive differenes between the seven hospitals seleted for the study sample and the remainin seven that were not. Nevertheless, the possibility of seletion bias would have been redued and the ase for eneralizability improved if all 14 hospitals (or a random sample from all 14 hospitals) had been studied. This study did not onsider the possible downstream benefits to either the health system or other departments in the partiipatin hospitals of havin better trained and more hihly enaed and motivated linial staff as a result of the QI intervention. Aordin to the literature, QI ollaboratives have produed downstream benefits suh as improved provider enaement with their work, and inreased trust amon providers (23); better staff relationships; and the extension of QI approahes to other areas within the failities (24). The hiher-quality linial are for diarrhea and pneumonia resultin from the PHI Collaborative intervention may also have led to a derease in morbidity and mortality from other auses, but these outomes were not measured. If these benefits did our and had been quantified and inluded in the analysis, the ost-effetiveness of the intervention would be hiher. The intervention s effet on the are of those dianosed with malnutrition (without pneumonia or diarrhea) was not onsidered either, even thouh part of the intervention addressed quality of are in this area. In addition, the authors did not onsider fators related to ost-effetiveness from the soietal perspetive, suh as potential ost savins for families arin for hildren with diarrhea and pneumonia after their dishare from the hospital. One aain, iven the dereased burden of disease assoiated with the intervention, the ost to families and areivers would have most likely dereased, improvin overall ost-effetiveness. Reommendation and onlusion Due to the improved patient outomes resultin from the intervention, overall hospital expenditures for the are of hildren with diarrhea and pneumonia 458 Rev Panam Salud Publia 30(5), 2011
Brouhton et al. Cost-effetiveness of improvin pediatri hospital are in Niaraua Oriinal researh would not inrease if MINSA expanded the intervention to other pediatri hospitals in Niaraua that were not part of the initial implementation. The $45 124 ost of the intervention, when divided by the number of hildren admitted to hospital pediatri wards with either diarrhea or pneumonia, is equivalent to about $3 per hild served. This is less than 0.7% of the averae ost of hospitalization for either of the two onditions. The derease in the overall burden of disease assoiated with the intervention indiates that it may lead to a substantial net benefit to the health system. Assumin that the ost of implementin the QI intervention in other pediatri failities is approximately the same as it was in the initial seven failities, the $3-per-hild investment may produe substantial ost-savins per DALY averted. Given the potential for lare ost-savins, implementation of the QI intervention by MINSA throuhout the ountry seems prudent. However, further researh is needed to establish whether or not the improvements seen were ompletely attributable to the intervention before the authors an unreservedly reommend its adoption to all hospitals in this settin. Aknowledments. The authors thank the staff and administrators of the hospitals for their partiipation in this study, and the Niarauan Ministry of Health (MINSA) for their support and assistane. This work was supported by the Amerian people throuh the United States Aeny for International Development (USAID) and its Quality Assurane Projet (QAP) (ontrat no. GHN-I-01-00003-00) and Health Care Improvement (HCI) Projet (ontrat no. GHN-I-01-07-00003-00), implemented by University Researh Co., LLC (URC). REFERENCES 1. World Health Oranization. Disease and injury ountry estimates: burden of disease. Deaths estimates for 2008 by ause for WHO Member States [Internet]. Geneva: WHO; 2009. Available from: http://www. who.int/healthinfo/lobal_burden_disease/ estimates_ountry/en/index.html. Aessed 8 Marh 2011. 2. Beker-Dreps S, Paniaua M, Dominik R, Cao H, Shah NK, Moran DR, et al. Chanes in hildhood diarrhea inidene in Niaraua followin 3 years of universal infant rotavirus immunization. Pediatr Infet Dis J. 2011;30(3):243 7. 3. Duke T. Child survival and IMCI: in need of sustained lobal support. Lanet. 2009; 374(9687):361 2. 4. USAID Health Care Improvement Projet. Improvin pediatri are in Niaraua. Bethesda, MD: University Researh Co., LLC; 2008. 5. Nembhard IM. Learnin and improvin in quality improvement ollaboratives: whih ollaborative features do partiipants value most? Health Serv Res. 2009;44(2 Pt 1): 359 78. 6. World Health Oranization; UNICEF. Interated Manaement of Childhood Illness hart booklet. Geneva: WHO; 2008. 7. World Health Oranization; UNICEF. Model IMCI handbook: interated manaement of hildhood illness. Geneva: WHO; 2005. 8. Exhanerateusd.om. 2006 historial exhane rates [Internet]. Available from: http://www.exhanerateusd.om/historial urreny/2006. Aessed 15 September 2010. 9. Murray C, Lopez A. The lobal burden of disease. Cambride, MA: Harvard University Press; 1996. 10. World Health Oranization. Global health observatory data repository: ountry statistis. Niaraua [database on the Internet]. Geneva: WHO; 2009. Available from: http:// apps.who.int/hodata/?vid=14800&theme= ountry. Aessed 8 Marh 2011. 11. Kim SY, Goldie SJ, Salomon JA. Costeffetiveness of Rotavirus vaination in Vietnam. BMC Publi Health. 2009;9(1):29. 12. De la Hoz F, Alvis N, Narváez J, Cediel N, Gamboa O, Velandia M. Potential epidemioloial and eonomial impat of two rotavirus vaines in Colombia. Vaine. 2010; 28(22):3856 64. 13. World Health Oranization. CHOosin Interventions that are Cost Effetive (WHO- CHOICE). Threshold values for intervention ost-effetiveness by reion [table on the Internet]. Geneva: WHO; 2010. Available from: http://www.who.int/hoie/osts/cer_ levels/en/index.html. Aessed 23 Auust 2010. 14. World Bank. World Development Indiators database. Gross national inome per apita, 2009: Atlas method and PPP [Internet]. Washinton: World Bank; 2010. Available from: http://siteresoures.worldbank.or/ DATASTATISTICS/Resoures/GNIPC.pdf. Aessed 10 Auust 2010. 15. Ilesias CP, Drummond MF, Rovira J; NEVALT Projet Group. Health-are deisionmakin proesses in Latin Ameria: problems and prospets for the use of eonomi evaluation. Int J Tehnol Assess Health Care. 2005;21(1):1 14. 16. Goldie SJ, Diaz M, Constenla D, Alvis N, Andrus JK, Kim SY. Mathematial models of ervial aner prevention in Latin Ameria and the Caribbean. Vaine. 2008;26 Suppl 11: L59 72. 17. Rheinans RD, Constenla D, Antil L, Innis BL, Breuer T. Potential ost-effetiveness of vaination for rotavirus astroenteritis in eiht Latin Amerian and Caribbean ountries. Rev Panam Salud Publia. 2007;21(4):205 16. 18. Centenari C, Gurel RQ, Bohland AK, Oliveira DM, Faraher B, Cuevas LE. Rota virus vaination in northeast Brazil: a laudable intervention, but an it lead to ost-savins? Vaine. 2010;28(25):4162 8. 19. Vilhez S, Reyes D, Paniaua M, Buardo F, Möllby R, Weintraub A. Prevalene of diarrhoeaeni Esherihia oli in hildren from León, Niaraua. J Med Mirobiol. 2009;58 (Pt 5):630 7. 20. Ministerio de Salud (NI). Sala de situaión. Boletín epidemiolóio. May 3. Manaua: MINSA; 2011. Available from: http://www. minsa.ob.ni/index.php?option=om_ ontent&view=artile&id=31:sala-desituaion&atid=6:direion-eneralviilania-de-la-salud-publia&itemid=40. Aessed 15 September 2011. 21. UNICEF. At a lane: Niaraua [Internet]. New York: UNICEF; 2010. Available from: http://www.unief.or/infobyountry/ niaraua.html. Aessed 26 Auust 2010. 22. den Enelsen C, van der Werf C, Matute AJ, Delado E, Shurink CA, Hoepelman AI. Infetious diseases and the use of antibiotis in outpatients at the emereny department of the University Hospital of León, Niaraua. Int J Infet Dis. 2009;13(3):349 54. 23. Ayers LR, Beyea SC, Godfrey MM, Harper DC, Nelson EC, Batalden PB. Quality improvement learnin ollaboratives. Qual Mana Health Care. 2005;14(4):234 47. 24. Bradley JE, Mayfield MV, Mehta MP, Rukone A. Partiipatory evaluation of reprodutive health are quality in developin ountries. So Si Med. 2002;55(2):269 82. Manusript reeived on 10 Otober 2010. Revised version aepted for publiation on 13 May 2011. Rev Panam Salud Publia 30(5), 2011 459
Oriinal researh Brouhton et al. Cost-effetiveness of improvin pediatri hospital are in Niaraua resumen Costo-efiaia de la mejora de la atenión hospitalaria pediátria en Niaraua Palabras lave Objetivo. Determinar el osto y la efiaia en funión del osto de una intervenión para mejorar la alidad de la atenión de niños on diarrea o neumonía en 14 hospitales de Niaraua, sobre la base de la informaión sobre astos y la mediión de las reperusiones. Métodos. Se ompilaron datos sobre la duraión de la hospitalizaión y la mortalidad de una muestra aleatoria de 1 294 historias línias ompiladas en 7 de los 14 hospitales partiipantes antes de la intervenión (2003) y 1 505 historias línias ompiladas después de dos años de ejeuión de la intervenión ( postintervenión, 2006). Los años de vida ajustados en funión de la disapaidad (AVAD) se obtuvieron de los resultados asisteniales. Se alularon los ostos de hospitalizaión seún los reistros de los hospitales y del Ministerio de Salud, y datos del setor privado. Los ostos de la intervenión se obtuvieron de los reistros ontables del proyeto. Para alular la relaión osto-efiaia inremental se usó un análisis de árbol de deisiones. Resultados. La duraión promedio de la hospitalizaión disminuyó de 3,87 y 4,23 días antes de la intervenión a 3,55 y 3,94 días después de la intervenión para la diarrea (P = 0,078) y la neumonía (P = 0,055), respetivamente. La letalidad disminuyó de 45/10 000 y 34/10 000 antes de la intervenión a 30/10 000 y 27/10 000 después de la intervenión para la diarrea (P = 0,062) y la neumonía (P = 0,37), respetivamente. Los ostos totales promedio de la hospitalizaión y de los antibiótios para ambos dianóstios fueron de US$ 451 (intervalo de onfianza [IC] de 95%: US$ 419 a US$ 482) antes de la intervenión y US$ 437 (IC 95%: US$ 402 US$ 464) después. La intervenión representó un ahorro de ostos en uanto a los AVAD (IC 95%: US$ 522 a US$ 32 por ada AVAD evitado) y ostó US$ 21 por ada día de hospitalizaión evitado (IC 95%: US$ 45 a US$2 04). Conlusiones. Después de dos años de ejeuión de la intervenión, la duraión de la hospitalizaión y la mortalidad por diarrea disminuyeron, junto on la duraión de la hospitalizaión para la neumonía, sin un aumento de los ostos de hospitalizaión. En aso de que estos ambios fueran totalmente atribuibles a la intervenión, esta representaría un ahorro de ostos. Costos de hospital; hospitales pediátrios; análisis osto-benefiio; diarrea; neumonía; prestaión de atenión de salud; Niaraua. 460 Rev Panam Salud Publia 30(5), 2011