Maintenance Versus Replacement of Medical Equipment in Nepal (2022) PDF

Hillebrecht et al. BMC Health Services Research (2022) 22:1023 https://doi.org/10.1186/s12913-022-08392-6 RESEARCH Open Access Maintenance versus re...

Hillebrecht et al. BMC Health Services Research (2022) 22:1023 https://doi.org/10.1186/s12913-022-08392-6 RESEARCH Open Access Maintenance versus replacement of medical equipment: a cost‑minimization analysis among district hospitals in Nepal Michael Hillebrecht1,2,3*, Constantin Schmidt1, Bhim Prasad Saptoka4,5, Josef Riha6, Matthias Nachtnebel7 and Till Bärnighausen1,8,9,10 Abstract Background: About half of all medical devices in low- and lower-middle-income countries are currently non-opera- tional because equipment maintenance is lacking. Thus, choosing a cost-efficient equipment maintenance approach has the potential to increase both the quantity and quality of important health services. Between 2010 and 2014 Nepal’s Ministry of Health chose two of its development regions to pilot the contracting-out of maintenance services to the private sector. We develop a cost model and employ different data to calculate the cost of this contracted-out scheme. The latter we compare with two additional common approaches to maintenance: in-house maintenance and no maintenance. Methods: We use invoiced pilot program costs, device depreciation estimates from the literature, and hospital case numbers from Nepal’s Health Management Information System. We estimate net-present values for a three-year horizon, incorporating both fixed and operational cost. Operational costs include downtime cost measured as lost revenues due to non-working equipment. Results: The contracted-out maintenance scheme shows a strong relative cost performance. Its cost after 3 years amount to 4,501,574 International Dollars Purchasing Power Parity (I$ PPP), only 90% of the cost with no maintenance. The contracted-out scheme incurs 670,288 I$ PPP and 3,765,360 I$ PPP in fixed cost and operational cost, respectively. The cost for replacing broken devices is 1,920,467 I$ PPP lower with maintenance. In addition, after 3 years total cost of contracted-out maintenance is 489,333 I$ PPP (11%) below total cost of decentralized in-house maintenance. After 10 years, contracted-out maintenance saves 2.5 million I$ PPP (18%) compared to no maintenance. Conclusions: We find that contracted-out maintenance provides cost-efficient medical equipment maintenance in a lower-middle income context. Our findings contrast with studies from high- and upper-middle-income countries, which reflect contexts with more in-house engineering expertise than in our study area. Since the per hospital fixed cost decrease with scheme size, our results lend support to an expansion of contracted-out maintenance to the remaining three development regions in Nepal. Keywords: Nepal, Maintenance, Health management, Health technology, Asset management, Costing study Background Modern medicine is unthinkable without medical *Correspondence: [email protected] devices, such as X-rays, ultrasound machines or micro- 1 Heidelberg Institute of Global Health (HIGH), Medical Faculty and University scopes. They facilitate the prevention, diagnosis, and Hospital, Heidelberg University, Heidelberg, Germany treatment of illnesses and as such directly affect human Full list of author information is available at the end of the article © The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. The Creative Commons Public Domain Dedication waiver (http://​creat​iveco​ mmons.​org/​publi​cdoma​in/​zero/1.​0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Hillebrecht et al. BMC Health Services Research (2022) 22:1023 Page 2 of 9 lives. At the same time, the acquisition and ownership Methods of medical devices is expensive – including considerable Pilot intervention cost of maintenance – and competes with other invest- In the fiscal year 2009/10, Nepal’s Ministry of Health ments targeted towards quality health care provision. operated 65 primary care hospitals (district hospitals) Choosing the right approach to manage and maintain and 27 higher level hospitals (zonal, regional, and central medical equipment thus might come with substantial hospitals). The public health system was financed efficiency and performance gains for a country’s health through the central state government budget, contribu- care system. However, comprehensive medical equip- tions at point of service – free services were provided ment maintenance policies are rare in the Global South to poor and vulnerable populations – and international [2, 3]. Challenges include the absence of funding for donors. Our study analyzes the pilot program of Nepal’s maintenance schemes among the international donor national contracted-out maintenance scheme for medi- community , miss-management, for instance, through cal equipment of medium complexity. The scheme was misallocation of spare parts or non-aligned purchases run by the Ministry of Health with financial support of medical devices [3, 4], as well as the lack of qualified from the German KfW Development Bank and techni- technical staff [4–6]. cal assistance from management4health, an international In addition, scientific evidence on the performance consultancy. In addition, researchers from Heidelberg and effects of medical equipment maintenance schemes University’s Institute for Global Health conducted an in low and lower-middle income countries is scarce. A independent ex-post review of how the program was set of studies investigates the current state of equipment implemented. We draw on data generated over the pro- maintenance by building up inventories and conducting gram’s pilot between 2010 and 2014 in Nepal’s Far-West- expert interviews to assess equipment operability and ern and Mid-Western region. As no inventory of medical qualitative maintenance management recommendations devices in public hospitals existed before the pilot, an [2, 5, 7–9]. Two studies assess the performance of medi- initial rapid inventory assessment was conducted. Based cal equipment schemes in an empirical fashion [6, 10]. on this assessment, the government’s Physical Assets Yet, to the best of our knowledge, the literature provides Management unit commissioned both preventive and no answers to the question whether medical equipment corrective maintenance services for all medical devices maintenance delivers long-term cost savings and quality of medium complexity located in public regional, sub- of care improvements from a health system perspective. regional, zonal, and district hospitals. Our study focuses Thus, our paper focuses on the cost dimension and on the 19 district hospitals which cover an average popu- aims to provide novel evidence on several research gaps. lation of 291,000 individuals, each. First, it provides the first detailed cost analysis of a con- In addition, the contractor’s service package covered tracted-out medical equipment maintenance scheme maintenance of cold chain equipment in district health covering multiple hospitals in a lower-middle income offices and equipment-related training to hospital staff. country context. Second, it includes the first – albeit Inventory management and procurement of new devices hypothetical - comparative cost efficiency assessment remained under the responsibility of the Ministry of of the three most common maintenance modes: con- Health. In total, the contractor engaged 14 employees tracted-out, in-house, and no maintenance. Last, this on this job (managing director, four biomedical engi- study makes a methodological contribution by including neers, six biomedical engineering technicians and three downtime cost into the equation. We use a simple cost- logistics managers), of which the majority (except the ing model with fixed, operational, and downtime cost to managing director, the chief engineer and one logistics estimate the net present value under these three different manager) was assigned to one of the two regional main- maintenance scenarios. tenance workshops, in Nepalgunj (Mid-West) and Dan- Nepal, the context of this study, provides a showcase for ghadhi (Far-West). Biomedical engineering technicians the successful stepwise adaption of a maintenance man- were trained at BMET Training Centre in Kathmandu agement system, covering medical equipment of medium and provided the first line of service, by moving between complexity (such as X-ray and ultrasound). Based on an workshops and hospital facilities as required. Biomedi- initial equipment inventory and external donor support, cal engineers provided backstopping. Job assignments the government delegated responsibility for equipment were decided by the respective workshop manager as maintenance services, as well as trainings for in-house per maintenance schedule, job urgency and expertise staff, to a private contractor. Before the adaptation of the required. maintenance system, non-operational devices were sim- During the mobilization phase, the contractor visited ply replaced with new equipment, which is still the reality all hospitals for a thorough inspection and the creation in many low-income countries. of a final list of devices. Maintenance activities over the Hillebrecht et al. BMC Health Services Research (2022) 22:1023 Page 3 of 9 contract period of three years included six semi-annual equipment-based health services from Nepal’s national maintenance visits for non-critical devices. Critical Social Health Security Program. devices (such as X-ray, operating lights, electrosurgical From within Nepal’s pilot program, we draw on two units, sterilizing units in higher-than-district-level hos- sources of information. First, we use records from pre- pitals) received three preventive maintenance visits per ventive and corrective maintenance activities and the year. In addition, the contract included the provision of number of devices from the program’s MMIS. As the corrective maintenance services, with different support inclusion of downtime cost requires the combination of response rates being offered for critical and non-critical MMIS and HMIS data, our merged dataset for analysis devices. To monitor maintenance activities and pay- covers six medical devices out of the total 15 covered in ments, the Physical Asset Management unit centrally the MMIS. For considerations of space and easier com- established a digital maintenance management informa- parability this study focuses on district hospitals, only. tion system (MMIS) prior to the mobilization phase. It Second, we include invoiced maintenance services and records all maintenance and repair activities through the other cost accounts from the pilot project’s implemen- continuous collection of so-called “jobcards” and thereby tation phase. provides an up-to-date device inventory. Table 1 shows descriptive statistics from our set of 19 district hospitals, for our selection of devices. In total we include 223 devices, or about 13 devices per hospital. Due Data to differences in hospital size, the number of devices var- Nepal’s Health Management Information System (HMIS) ies greatly across hospitals. Similarly, there are large dif- collects a wide range of health provider information on ferences in equipment values. In total, all 53 centrifuges a regular basis. It reports monthly data on patient num- cost about 47,000 International Dollar Purchasing Power bers, services delivered, and performance indicators for Parity (I$ PPP) while the sum of all 47 X-rays amounts all public health facilities. The data management unit of to about 2.25 million I$ PPP. Laboratory equipment use the Ministry of Health gathers and publishes HMIS data is frequent with around 56 services per 1000 inhabitants to facilitate administrative procedures and to inform per year. ECG, in contrast only provides 0.2 services per policy decisions. Our data contain case numbers of six 1000 inhabitants. On average, each device required six equipment-based health services, which we extract for corrective maintenance visits per year. the two reporting rounds 2008/09 and 2014/15 (Table 1). To put HMIS headcounts into perspective, we relate dis- Cost equation trict hospital case numbers to the population of respec- Taking a health system planner perspective, our cost tive catchment areas. For medical equipment downtime evaluation is a comparative assessment of three mainte- cost calculation we obtained remuneration rates for nance scenarios (contracted-out maintenance, in-house Table 1 Summary statistics of device types included in the analysis 1 2 3 4 5 6 a Devices total Devices per Total value Number of S ­ ervices Corrective Device Lifetime HF (I$ [PPP], Maintenance° 2016) Mean SD 1 Laboratory equipment a Analyzer 11 0.6 1.13 292,169 55.64 5 7.5 b Centrifuge 53 2.9 1.08 125,785 (Analyzed in 41 9.0 c Microscope 41 2.3 1.07 311,377 ­laboratoryb) 19 10.0 2 Monitoring a ECG 21 1.2 0.79 247,203 0.19 14 8.0 3 Ultrasound imaging 50 2.8 1.07 2,230,901 5.39 9 10.0 4 X-ray imaging 47 2.6 1.54 6,065,638 9.06 73 8.0 Total 223 12.4 9,273,072 70.28 161 Notes: Numbers for hospitals included into our costing study. Abbreviations: ecg, electrocardiogram. aPer 1,000 inhabitants, per year. Average values over 2010/11 till 2013/14 Health Management Information System rounds. bNumber of cases examined in the laboratory using one or more of analyzer, centrifuge, or microscope. °Total number of corrective maintenance visits over six rounds. Data sources: Devices total and Devices per HF, Health Management Information System; Total value, Consultant; Number of Services and Corrective Maintenance, HMIS; Device Lifetime, engineering literature [12, 13] Hillebrecht et al. BMC Health Services Research (2022) 22:1023 Page 4 of 9 maintenance, and no maintenance) including cost at the approach of downtime calculation we provide rather hospital and central management level. We employ a lower bound estimates for the true economic cost related costing model with fixed and operating costs, as set out to an increase in equipment downtime. in eq. (1). Fixed costs are composed of establishing the We define equipment downtime as the number of central administrative unit and either building in-house days a device is unavailable for services due to technical workshops or setting up the contracted-out scheme. defects. We calculate downtime from the time passed Operating costs accrue for central maintenance system between the request for a corrective maintenance task management, device replacements and maintenance (as recorded in the MMIS) and task completion. Since it operations. Our model also accounts for downtime as is very unlikely that a request is filed immediately after an additional operational cost component, to control for the device breaks down, we consider 15% of the time lost hospital revenues due to a lower share of operational passed between a corrective maintenance job request and devices. Our outcome indicator is the net present value the equipment’s last inspection, whether it was sched- of total cost during the length of the pilot (three years), uled or not, as additional downtime. We used downtime for which we assume a discount rate (r) of 5%. Prices were from the year before the maintenance pilot was started adjusted for inflation and local currencies, and converted as a benchmark for downtime under no maintenance. We into I$ PPP using the World Bank’s World Development provide a robustness check for this assumption in Table Indicators. Please see in Table A9 in the appendix for A5 in the Appendix. From HMIS reports we then extract detailed information. the number of services provided by each device type in T Ct Management + Ct Replacement + Ct Maintenance + Ct Downtime C = C Fixed +C Operating = C Fixed + (1) (1 + r)t t=1 Cost components an average-sized district hospital per day. Next, we assign Replacement cost is the average expected investment a price tag to each service by employing health services in new devices based on purchasing value and number remuneration rates from Nepal’s National Social Health of devices. For instance, if a device has a lifetime of 10 Security Program. These numbers reflect rates charged years, the hospital should expect average yearly replace- by public hospitals and, depending on the service, range ment costs of one tenth the total purchasing value of its from 3.0 to 13.5 I$ PPP per service. As such, they provide current stock of this device. A cross-country delphi study a lower bound of the health services’ true market value. provides lifespan intervals for four different equipment By multiplying the number of days a device is unavail- categories, distinguishing between good and bad quality able with the number of services per day and renumera- devices, as well as between good and bad maintenance tion rates, we estimate the annual downtime cost for each. Based on these intervals we have calculated an aver- device. Table A1, in the Appendix, provides a definition age lifetime reduction rate incurred by applying bad (or of each cost type and the respective data source. no) maintenance instead of good maintenance, amount- ing to 39%. Cost scenarios To the best of our knowledge, downtime cost has not The benchmark scenario for our study is contracted- been included into costing studies of medical equip- out maintenance, which was piloted in our study area. ment maintenance, yet, even if they might constitute Accordingly, cost estimates mainly come from invoiced a substantial share of overall equipment-related cost program costs and additional figures from the litera-. At the hospital level, the cost of equipment down- ture (column 2 in Table 2). Due to mentioned data con- time are missed profits due to fewer equipment-related straints, we had to narrow down the analysis to six device services offered and reimbursed. From a health system types. Therefore, we reduce IT, transport, and coordina- perspective, additional costs may accrue as non-func- tion components of management cost accordingly. We tional equipment will result in inadequate diagnosis and keep other management cost the same, like the number therapy, thereby hampering quality of care and poten- of rooms and the initial inventory. We split the con- tially harming patients. A program’s ability to reduce tracted-out cost of the mobilization phase and the pilot equipment downtime thus may, by itself, provide a use- by number of devices. Table A2 in the Appendix provides ful hospital performance benchmark measure for com- a detailed description of the assumptions we made. parison. We therefore deliberately present and discuss In the hypothetical no maintenance scenario, we downtime cost differences across maintenance modes. assume broken devices to be exchanged for new devices However, we are aware that with our hospital-focused when a fault is detected (column 1 in Table 2). This Hillebrecht et al. BMC Health Services Research (2022) 22:1023 Page 5 of 9 Table 2 Maintenance scenarios and corresponding program cost Cost scenario: A B C No maintenance Contracted-out maintenance: Intervention In-house maintenance Assumed equipment lifetime reduction: 39% 0% 0% 0% (1) (2) (3) (4) Fixed Cost (in t = 0) Management, contracted   IT-System – 181,196 181,196 181,196   Transport – 71,199 71,199 71,199   Office Space – 6875 6875 6875 Investment, in-house   Workshops – – – 421,830   Tools – – – 118,754 Preventive maintenance, contracted    Rapid inventory assessment – 251,355 251,355 –   Mobilization phase – 159,663 159,663 – Total fixed cost 0 670,288 670,288 799,855 Operating Cost (in t = 1,2,3) Management, contracteda   IT-System – 71,094 71,094 71,094   Car – 17,977 17,977 17,977   Annual salaries 178,720 178,720 178,720 Replacement cost 4,924,275 3,003,808 3,003,808 3,003,808 Maintenance cost    Invoiced program cost, contracted    Preventive maintenance – 465,777 – –    Spare parts – 27,914 27,985 –    Calculated program cost, contracted    Preventive maintenance – – 72,168 –    Corrective maintenance – – 14,286 –    Calculated program cost, in-house    Annual salaries – – – 288,856     Spare parts and tests – – – 185,349 Total operating cost 4,924,275 3,765,360 3,386,037 3,745,804 Downtime Cost Total downtime costb 104,547 65,925 not applicable Total program cost (NPV) 5,028,823 4,501,574 4,056,325 4,545,658 Notes: Each cell reports net present value of total cost in 2016 I$ PPP over three years, calculated for the sum of 19 district hospitals from the two pilot regions. aAs maintenance management responsibilities do not include the procurement of new equipment, we do not include it in the management cost estimates. bAs we have no reliable estimate for downtime cost under inhouse maintenance, it is only considered for the comparison of replacement with contracting out scenarios. We assume downtime cost to be equal between full in-house and contracted out maintenance scenario well reflects the situation in Nepalese health Such a decentralized scheme describes a common com- facilities prior to the pilot. As such a scenario does not parison scenario. It rests on three important assump- involve maintenance activities, its main cost driver is tions. First, we assume that fixed and operational central more frequent device replacement. management costs are overall the same for contracted- The hypothetical in-house maintenance scenario del- out and in-house maintenance. In-house maintenance egates maintenance activities to the hospitals, while requires less central planning than contracted-out keeping up an administrative unit for monitoring and maintenance, but smaller performance incentives might steering at the federal ministry (column 4 in Table 2). increase the need for oversight. Second, we assume that Hillebrecht et al. BMC Health Services Research (2022) 22:1023 Page 6 of 9 both contracted-out and in-house maintenance perform much higher cost of salaries and spare parts of in-house equally well in keeping equipment downtime low. Third, maintenance. we assume that there is sufficient biomedical technical expertise available in the country to employ at least one Additional results biomedical engineer per district hospital. For salaries as To put the total numbers in context, we break down cost well as costs of spare parts, tests, and facilities, we rely to the hospital level. To this end we assume that all fixed on expert estimates from Nepal’s Physical Asset Man- and operational costs are equally shared by the number agement unit (Table A3 in the Appendix). To allow for a of district hospitals and accordingly divide total program fair comparison, we recalculate operational maintenance costs by the number of 19 district hospitals. Again, we cost of Nepal’s contracted-out maintenance scheme (see consider the net-present value over a three-years period. column 3 in Table 2). Instead of drawing on invoiced According to Table A4 in the Appendix, total per hospi- maintenance cost only, where training of in-house tech- tal cost for a no maintenance approach and contracted- nicians makes up a considerable share, we use corrective out maintenance amount to about 264,675 I$ PPP and maintenance entries from the MMIS and assume a rea- 236,925 I$ PPP, respectively. Assuming hospitals do not sonable flat rate for every single scheduled maintenance have to contribute to neither fixed nor operational man- inspection. agement cost, per hospital cost are 264,675 I$ PPP and 209,185 I$ PPP, for no maintenance and contracted-out maintenance, respectively. Results Calculating program cost for a longer time horizon No versus contracted‑out maintenance provides a second extension to our main analysis. Fig- Nepal’s contracted-out scheme requires initial invest- ure 1 illustrates how total equipment-related program ments of 670,288 I$ PPP for technical and management cost for a no maintenance and contracted-out scheme infrastructure. After three years, operational mainte- evolve over a ten-years period. After the first year, no nance cost – for both preventive and corrective mainte- maintenance is 223,140 I$ PPP cheaper than contracted- nance – amount to 493,762 I$ PPP (11.0% of total cost). out maintenance. For subsequent years, full replacement Operational management and downtime cost account costs accumulate disproportionally fast and after 10 for 267,790 I$ PPP (5.9% of total cost) and 65,925 I$ years exceed total contracted-out maintenance costs by PPP (1.5% of total cost), respectively. Together, these 2,529,639 I$ PPP (17.7%). One key result of this extended cost items add up to almost 1.5 million I$ PPP, which is analysis is that the additional investments of contracted- still less than half of the replacement cost (3 million I$ out maintenance already breaks-even after two years. [PPP]). In the absence of any maintenance (scenario A in Table 2), replacement cost amounts to 4,924,275 I$ PPP. In sum, this substantial cost difference in replacement Discussion cost leads to a cost advantage of contracted-out main- For our study setting in Nepal, contracted-out main- tenance over a no maintenance approach by 527,249 I$ tenance shows the highest cost efficiency across three PPP, after three years. Please note, that this comparative analyzed medical equipment maintenances modes. Its result mainly rests on the 39% lifetime reduction rate we cost advantage mainly stems from lower operating cost, have assumed. Our robustness check in Fig. A1 in the thus, the cost advantage of contracted-out maintenance Appendix shows that contracted-out maintenance out- increases with program duration. Similarly, contracted- performs no maintenance in a three-year time horizon as out maintenance’s relative cost advantage likely increases long as the lifetime reduction rate is above 30%. further when geographically upscaling the program. In our study we emphasize the importance of incorpo- rating downtime cost into the analysis when comparing Contracted‑out versus in‑house maintenance scenarios with different outcomes. The empirical Fixed and operational cost of contracted-out mainte- literature on medical equipment management is scarce nance are 129,566 I$ PPP (16.2%) and 359,767 I$ PPP and there is no established benchmark for measuring (9.6%) less than for in-house maintenance after three the benefits from equipment downtime reduction. In years. In-house maintenance includes building a work- a first attempt, we approximate the disutility of down- shop and equipping it with tools, accumulating to about time by missed hospital revenue. We find contracted-out 799,855 I$ PPP. Maintenance cost includes technical in- maintenance to be effective in reducing downtime cost house staff and requires hardware costing about 474,205 when compared with no maintenance. Even if downtime I$ PPP. Thus, the cost difference between contracted- cost from missed equipment-based revenues amount to out and in-house maintenance is mainly driven by the less than 2% of total cost it may serve as an important benchmark outcome by itself: minimizing equipment Hillebrecht et al. BMC Health Services Research (2022) 22:1023 Page 7 of 9 Fig. 1 Comparative cost assessment for a ten-year time horizon. Notes: Abbreviations: mio. I$, Million International Dollars (Purchasing Power Parity) downtime to improve quality of care and reduce treat- address the high data requirements needed for such an ment-induced health risks seems to be a worthwhile exercise. Specifically, we include detailed information goal from a health systems planner’s perspective. In this on device asset values and lifetimes, maintenance activi- regard Nepal provides a relevant context. In the absence ties, equipment-related service quantities and prices for of any maintenance scheme Nepali public hospitals were 19 district hospitals and six medical devices. This com- unable to fix even minor equipment malfunctions and pares favorably to existing empirical studies of medical suffered from a lack of equipment-related staff train- equipment maintenance, which consider not more than ing. As a large share of complex medical equipment two hospitals or investigate only incremental changes to in resource-constrained settings is donated through existing maintenance programs [20–23]. bilateral or multilateral development cooperation, our Our findings on the comparative advantage of con- insights are relevant for both policy makers and funding tracted-out over in-house maintenance contradict simi- institutions. In-house maintenance is a common and lar studies within an university hospital setting in the thus policy relevant scenario. Our calculations provide United States and Brazil [21, 22]. Both these studies a first cautious attempt to include such a scenario into a favor in-house maintenance over contracted-out main- comparative analysis. Nevertheless, the cost items rely on tenance. The reason for these different results might be hypothetical measures from Nepal’s Physical Asset Man- that engineering expertise is much more available in uni- agement unit and should be interpreted respectively. versity hospitals of high and upper-middle-income coun- This study is the first comprehensive costing study of tries than in public hospitals of lower-middle-income medical equipment maintenance in a resource-con- countries. strained country. The quantitative literature on medical It is important to note the following limitations of equipment in low- and lower-middle-income countries our study. First, we were not able to directly measure is very limited; most studies either provide simple equip- the difference in lifetime between well-maintained and ment inventories or purely qualitative attempts to explore non-maintained devices. As often the case in resource- maintenance practices [5, 7–9]. However, many low- constrained settings, Nepal has established an asset income countries lack a proper medical equipment management inventory on medical equipment very maintenance policy and the capability to implement any recently and the pilot was too short to track changes in of the literature’s recommendations [18, 19]. By taking a equipment lifetime. To this end, we have to rely on comprehensive approach and including no maintenance estimates from the published literature. Thus, the as a benchmark, this paper seeks to include this reality 39% lifetime reduction rate should be taken with cau- into the analysis. A favorable data situation enables us to tion. As our robustness check reveals, contracted-out Hillebrecht et al. BMC Health Services Research (2022) 22:1023 Page 8 of 9 maintenance’s relative cost advantage holds for an Abbreviations HMIS: Health Management Information System; MMIS: Maintenance manage- assumed equipment lifetime reduction greater than ment information system; I$ (PPP): International Dollar (Purchasing Power 30% (Fig. A1 in the Appendix). Second, we cannot Parity). directly observe how much time passed between the breakdown of a device and the malfunction being Supplementary Information reported to the maintenance support unit. Hence, we The online version contains supplementary material available at https://​doi.​ consider the time passed between a corrective main- org/​10.​1186/​s12913-​022-​08392-6. tenance job request and the equipment’s last inspec- Additional file 1. Data sources, assumptions, and robustness checks. tion and assume that 15% of such a time window is Information on data sources (Tables A1, A6 & A7), assumptions (Table A2 effective equipment downtime. Changing the time lag & A3), robustness checks (Fig. A1 & Table A5), and additional results (Tables to 5% or 25% of the time passed since the last inspec- A4 & A8). tion does not alter our qualitative results (Table A5 in Additional file 2. Full cost calculations and formatted data. Detailed steps the Appendix). Third, it’s important to note that data of the cost calculation and all data needed for replication in pre-processed format. quality from management information systems may vary substantially, depending on the context, includ- Acknowledgements ing underlying data entering procedures, built-in data Not applicable. validations, incentives for misreporting and others. We conducted careful quality and plausibility checks with Authors’ contributions MH designed the study and the costing model. MH and CS analyzed the data, our data from Nepal’s Health Management Informa- interpreted the results, and drafted the manuscript. TB supervised the project tion System (HMIS) and Maintenance Management and gave important input at all stages. All authors collected data, critically Information System (MMIS) and are confident that revised the manuscript, and approved the final version for publication. they do not draw a skewed picture of the reality. How- Funding ever, we want to emphasize the potential inherent limi- Open Access funding enabled and organized by Projekt DEAL. The KfW tations that arise by using management information funded the initial independent evaluation of the maintenance pilot carried out at the Heidelberg Institute of Global Health. The KfW had no role in the system data. design or execution of the present study or the writing of this paper. As with many empirical studies, a final limitation is that some of our results might not be easily applicable Availability of data and materials The dataset supporting the conclusions of this article is included within the to other circumstances. The pilot catchment area cov- article (and its additional file(s)) with exception of unformatted maintenance ers only about seven million people, roughly one quar- job-cards. Job-card data is available from the corresponding author on reason- ter of Nepal’s total population. Due to its relatively able request. low operational cost, we expect contracted-out main- tenance to perform even better when being scaled up Declarations to the country level. This motivates the cost evaluation Ethics approval and consent to participate of Nepal’s country-wide contracted-out maintenance Not applicable. scheme, which has been implemented in 2016, for Consent for publication future research. It is worth noting that Nepal’s terrain, Not applicable. with many hard-to-reach areas, may constitute a case where maintenance cost is relatively high compared to Competing interests MH, CS and TB declare that they have no competing interests. BS declares that other countries. he is an employee at the Ministry of Health in Nepal (MoH). He contributed to this study in his role as a PhD student at Ludwig-Maximilian University Munich, Germany, and independent of his professional engagement. SR declares that GOPA Consulting Group (back then: EPOS Health Management), on behalf Conclusion of SR, received honoraria from KfW Development bank to support the pilot To summarize, it is remarkable that contracted-out main- project implementation in Nepal. MN declares that he has been an employee tenance has by far the lowest cost, even when initial at KfW Development Bank during the time of the study. The authors are solely responsible for the opinions, hypotheses and conclusions or recommenda- investments are considered, while reducing equipment tions expressed in this publication, and they do not necessarily reflect KfW’s downtime. Thus, this study confirms contracted-out and MoH Nepal’s vision. maintenance as a cost-efficient tool to increase equip- Author details ment availability in health facilities for resource-con- 1 Heidelberg Institute of Global Health (HIGH), Medical Faculty and University strained settings. Future studies would benefit from using Hospital, Heidelberg University, Heidelberg, Germany. 2 South-Asia Institute, country-specific databases on equipment lifetime. In Heidelberg University, Heidelberg, Germany. 3 Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH, Eschborn, Germany. 4 Health addition, further research may widen the scope by look- Coordination Division, Ministry of Health and Population, Kathmandu, Nepal. ing into the role of in-house staff training and the central 5 Division of Infectious Diseases and Tropical Medicine, Center for International Physical Assets Management Policy. Health, University Hospital Munich, Ludwig-Maximilians-University, Munich, Hillebrecht et al. BMC Health Services Research (2022) 22:1023 Page 9 of 9 Germany. 6 Management4health, Frankfurt, Germany. 7 Health and Social 22. Da Rocha LS, Sloane EB, Bassani JWM. Optimal medical equipment main- Protection Asia, KfW Development Bank, Frankfurt am Main, Germany. tenance service proposal decision support system combining Activity 8 Africa Health Research Institute (AHRI), Somkhele, KwaZulu‑Natal, South Based Costing (ABC) and the Analytic Hierarchy Process (AHP). In: 2005 Africa. 9 MRC/Wits Rural Public Health and Health Transitions Research Unit IEEE engineering in medicine and biology 27th annual conference. IEEE; (Agincourt), School of Public Health, Faculty of Health Sciences, University 2006. p. 7103–6. of the Witwatersrand, Johannesburg, South Africa. 10 Center for Population 23. Miguel-Cruz A, Haugan GL, Rios-Rincón AM. The effects of asset specific- and Development Studies, Harvard University, Cambridge, MA, USA. ity on maintenance financial performance: an empirical application of transaction cost theory to the medical device maintenance field. Eur J Received: 30 January 2022 Accepted: 21 July 2022 Oper Res. 2014;237(3):1037–53. 24. Griffiths UK, Legood R, Pitt C. Comparison of economic evaluation meth- ods across low-income, middle-income and high-income countries: what are the differences and why? Health Econ. 2016;25:29–41. 25. Miguel-Cruz A, Rios-Rincón AM. Medical device maintenance outsourc- References ing: have operation management research and management theories 1. World Health Organization. Global atlas of medical devices. WHO Medical forgotten the medical engineering community? A mapping review. Eur J device technical series. Geneva: WHO Press; 2017. Oper Res. 2012;221(1):186–97. 2. Fonjungo PN, Kebede Y, Messele T, Ayana G, Tibesso G, Abebe A, et al. Laboratory equipment maintenance: a critical bottleneck for strength- ening health systems in sub-Saharan Africa? J Public Health Policy. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in pub- 2012;33(1):34–45. lished maps and institutional affiliations. 3. Reichenvater H, Matias LD. Is Africa a ‘graveyard’ for linear accelerators? Clin Oncol. 2016;28(12):e179–83. 4. Marks IH, Thomas H, Bakhet M, Fitzgerald E. Medical equipment donation in low-resource settings: a review of the literature and guidelines for surgery and anaesthesia in low-income and middle-income countries. BMJ Glob Health. 2019;4(5):1–9. 5. Ogembo-Kachieng’a M, Ogara WO. Strategic management of technol- ogy in public health sector in Kenya and South Africa. East Afr Med J. 2004;81(6):279–86. 6. Shah MT, Bhattarai S, Lamichhane N, Joshi A, LaBarre P, Joshipura M, et al. Assessment of the availability of technology for trauma care in Nepal. Injury. 2015;46(9):1712–9. 7. Alikhani P, Ganji H, Abtahi M, Vesal S, Naghdi B. Preventive maintenance of medical equipment in Alzahra Hospital, Isfahan, Iran 2013. Int J Health Syst Disaster Manag. 2013;1(4):217. 8. Maranga S, Kihiu J, Mutia D. Maintenance management of medical equip- ment in hospitals. Indust Eng Lett. 2012;2(3):9–19. 9. Bahreini R, Doshmangir L, Imani A. Influential factors on medical equip- ment maintenance management: in search of a framework. J Qual Maint Eng. 2019;25(1):128–43. 10. Ben HZ, Masmoudi M, Al HA, Khatrouch I, Masmoudi F. Quantitative techniques for medical equipment maintenance management. European J of. Ind Eng. 2016;10(6). 11. Department of Health Services. Annu Rep 2066/67 (2009/2010). 2010. 12. Social Health Security Committee. Social health security program - annual report. Teku, Kathmandu, Nepal; 2016. 13. World Bank. World Development Indicators [Internet]. 2021 [cited 2021 Mar 19]. Available from: https://​datat​opics.​world​bank.​org/​world-​devel​ opment-​indic​ators/ 14. Halbwachs H. Maintenance and the life expectancy of healthcare equip- ment in developing economies. Health Estate. 2000;54(2):26–31. 15. Dunscombe P, Roberts G, Valiquette L. Preventative maintenance and unscheduled downtime from an economic perspective. J Appl Clin Med Phys. 2000;1(2):68–75. 16. Jamshidi A, Rahimi S, Ait-kadi D, Bartolome A. Medical devices inspection and maintenance; a literature review. In: IIE annual conference proceed- ings. Institute of Industrial and Systems Engineers (IISE); 2014. Ready to submit your research ? Choose BMC and benefit from: 17. Edwards RT, McIntosh E. Applied health economics for public health practice and research. Oxford: Oxford University Press; 2019. fast, convenient online submission 18. Webber CM, Martínez-Gálvez G, Higuita ML, Ben-Abraham EI, Berry BM, Porras MAG, et al. Developing strategies for sustainable medical equip- thorough peer review by experienced researchers in your field ment maintenance in under-resourced settings. Annals of. Glob Health. rapid publication on acceptance 2020;86(1). support for research data, including large and complex data types 19. World Health Organization. Medical device donations: considerations for solicitation and provision. Geneva: WHO Medical device technical series, gold Open Access which fosters wider collaboration and increased citations WHO Press; 2011. maximum visibility for your research: over 100M website views per year 20. Nisreen H, Salloom A, Omer N. Medical devices service life cycle cost Management in Al Karak Hospital as a case study. J Account Market. At BMC, research is always in progress. 2015;4(2). 21. Rossi RP. Equipment maintenance: a cost comparison of vendor and Learn more biomedcentral.com/submissions inhouse service. Radiol Manag. 1989;11(4):28–34.