Evaluations of training and education interventions for improved infectious disease management in low-income and middle-income countries: a systematic literature review.

OBJECTIVES: To identify most vital input and outcome parameters required for evaluations of training and education interventions aimed at addressing infectious diseases in low-income and middle-income countries.
DESIGN: Systematic review. DATA SOURCES: PubMed/Medline, Web of Science and Scopus were searched for eligible studies between January 2000 and November 2021. STUDY SELECTION: Health economic and health-outcome studies on infectious diseases covering an education or training intervention in low-income and middle-income countries were included.
RESULTS: A total of 59 eligible studies covering training or education interventions for infectious diseases were found; infectious diseases were categorised as acute febrile infections (AFI), non-AFI and other non-acute infections. With regard to input parameters, the costs (direct and indirect) were most often reported. As outcome parameters, five categories were most often reported including final health outcomes, intermediate health outcomes, cost outcomes, prescription outcomes and health economic outcomes. Studies showed a wide range of per category variables included and a general lack of uniformity across studies.
CONCLUSIONS: Further standardisation is needed on the relevant input and outcome parameters in this field. A more standardised approach would improve generalisability and comparability of results and allow policy-makers to make better informed decisions on the most effective and cost-effective interventions. © Author(s) (or their employer(s)) 2022. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.

This is the first review (to our knowledge) to systematically assess health economic and health-outcome literature of training or education interventions on input and outcome parameters used for improved management of infectious diseases.

This review covers a wide variety of infectious diseases, allowing for comparisons across disease areas but also introducing high heterogeneity of results.

This study is prone to publication bias as it includes only data from published literature.

Introduction

Infectious diseases continue to be a major health challenge worldwide, with the highest burden in low-income and middle-income countries (LMICs).1 Over the past decades, improvements have been made in the management of infectious diseases by, among others, the introduction of widespread vaccine programmes,2 health programmes on malaria,3 HIV prevention4 and the widespread use of antimicrobials for bacterial infections.5 As a downside, widespread overuse of antimicrobials (among others) for treatment of infectious diseases has resulted in an increase of antimicrobial resistance (AMR) which could make future infections difficult or impossible to treat. Thus, to further reduce the global burden of infectious diseases, there is a need for (new) effective strategies that can be implemented at high speed with high coverage levels.6 These strategies should enable effective management of infectious diseases but also limit inappropriate use of antimicrobials to prevent further increase of AMR.

A variety of programmes have been implemented to address the management of specific diseases such as HIV, malaria or tuberculosis (TB)7 or the prescription of antimicrobials.8 Across the different disease programmes, commonalities can be found on two major topics. First, the implementation of diagnostics is an often used strategy across programmes, such as rapid diagnostic tests (RDTs) for malaria diagnosis9 or home-based testing for HIV detection.10 11 Second, education or training interventions are used across different infectious disease programmes. For example, physicians are trained and educated on improved prescription of antimicrobials,8 patients are taught about the importance of treatment adherence for antiretroviral therapy (ART)12 and individuals are informed on preventive measures that can be taken to prevent HIV or malaria infections.13 Evidently, there are similarities in the approaches that are used by the different programmes, but within a programme the interventions are often focused on one specific disease (eg, malaria, HIV). Hence, with finite financial resources, a decision needs to be made by policy-makers on a limited number of disease-specific programmes that can be incorporated in national health policy.

Policy-makers are informed by health economic analyses to maximise the impact on health and equity. The health economic impact is often expressed in costs per quality-adjusted life year gained (cost per QALY) or cost per disability-adjusted life year averted (cost per DALY), both of which combine morbidity and mortality (ie, quality and length of life).14 QALYs are predominantly used in higher income countries and DALYs in global health studies.15 Expressing health economic impact in cost per QALY or cost per DALY allows for comparing different health interventions across diseases.16

There are no consistent guidelines with input parameters and outcomes to report on in health economic evaluations of infectious disease interventions in LMICs.17 18 To close this gap, previous endeavours have been undertaken by the VALUE-Dx consortium to review health economic assessments of diagnostic interventions for infectious diseases.19 One of the conclusions of this consortium was that there is a lack of universal outcomes in the assessment of diagnostics. Parameter categories that were found across a multitude of studies included final health outcomes (QALY, DALY), antibiotic consumption and diagnostic test performance. This provides valuable insight in parameters to use for the health economic assessment of diagnostics. However, to our knowledge, comparable research is lacking on educational or training interventions for improved management of infectious diseases.

It is important to get a better understanding of input parameters and outcomes that have been used previously to guide future research efforts, to improve the quality of health economic assessments as well as the generalisability of results. Such guidance would specifically be relevant for LMICs, where the need for improved management of infectious diseases is most urgent,20 21 where health economic frameworks are less formalised, and where limitations are encountered in applying results from health economic studies into policy-making.22 Therefore, the objective of this review is to close the knowledge gap by identifying input parameters and outcomes reported in health economic and health-outcome studies on training or education interventions for infectious diseases in LMICs.

Methods

Search strategy

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines23 were used for this study (online supplemental appendix A). A systematic search of databases was performed, including PubMed/Medline, Web of Science and Scopus. The detailed search strategy per database can be found in online supplemental appendix B. Five queries were combined in the main query, which aimed to include studies that matched the following elements:

Population: individuals in LMICs (ie, countries and territories that are eligible to receive official development assistance as per the Organisation for Economic Co-operation and Development).24

Intervention: programmes that include an education or training intervention.

Disease focus: infectious diseases.

Type of research: health economic and health-outcomes articles.

Time period: January 2000–November 2021.

Duplicate articles were removed after which the title and abstract were scanned independently by two researchers (PWMvD and ADIvA). Full-text analysis was performed on potentially relevant articles.

Study selection

We included studies which, based on full text analysis, met the following inclusion criteria: (1) assessing the impact of either a training or education intervention; (2) focused on infectious diseases; (3) in LMICs; (4) in humans; and (5) reporting the impact of the intervention in either health or health economic outcomes. Studies were excluded if no intervention was applied (eg, review, protocol, cross-sectional or descriptive study), if the intervention did not include a training or educational aspect, in case the training was merely focused on the introduction of RDTs as test-and-treat strategy (which was the scope of the Value Dx consortium), and if the full text was not available or not available in English.

Data extraction

Included studies were systematically analysed and documented using a digital form (Google Forms; see online supplemental appendix C). Within the digital form, a distinction was made between health economic articles and health-outcomes articles. For health economic articles, a total of 57 variables were listed for data extraction, using the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) checklist as a basis.25 A total of 23 variables were listed for health-outcome articles. Variables captured were related to study design, disease focus, interventions, input parameters and outcomes.

Categorisation of results

To structure the findings of the review, a categorisation of the infectious diseases was made between acute febrile infections (AFI) (fever for <7 days), non-AFI (fever for >7 days)26 and other infectious diseases that are not primarily febrile. This categorisation is used throughout the results section, which consists of the following three subsections: interventions identified, input parameters identified and outcomes identified. Further breakdown of the results in each subsection is explained below.

For the training and education interventions that were found in the review, further clarity was given by positioning the different interventions on the healthcare spectrum, for which the definition from O’Connell et al27 was used. The interventions were positioned in four distinct phases, including (1) promotion of health, (2) prevention of developing a disease, (3) treatment, including patient identification and start of the treatment and (4) maintenance/postintervention care, which includes patient compliance in long-term care and provision of after care.27

Input parameters found were categorised into four categories. The first category was costs which entailed all cost parameters that were used to calculate a final cost outcome (eg, cost of medication, cost of personnel). The second category was defined as aetiology-specific characteristics, covering disease-specific parameters that could impact other parameters (eg, average duration of a disease to calculate QALYs or DALYs). The third category was population background, defined as population-related parameters that could impact other input or outcome parameters (eg, per cent of population at risk in a country). The fourth and final category consisted of intervention details, which put the intervention in a broader perspective (eg, percentage of individuals at risk targeted by the intervention).

Outcome parameters were also categorised, in nine separate categories. The first two categories were related to health effects, in which the distinction between final and intermediate outcomes was made. Final health outcomes were defined as a quantification of the health effect of an intervention, reported in a final outcome for a health (status) change (eg, death, QALYs, DALYs). Intermediate health outcomes were quantified as a change in a clinical indicator that might or might not lead to final health outcomes.28 The third category was defined as cost outcomes, which included parameters that reported the cost outcomes of a whole programme or a single intervention. The fourth category was defined as prescription outcomes, which included parameters that quantify the prescription practices like doses and frequency, often described in standardised units like the defined daily doses (DDD). The fifth category, health economic outcomes, entailed outcomes that were reported as incremental cost per unit of outcome, indicating the cost-effectiveness of an intervention (ie, cost per QALY). The sixth category was defined as behavioural outcomes, indicating the effect of an intervention on the behaviour of the targeted individual. The seventh category consisted of time-related outcomes, which included outcomes that indicated important time-related aspects as a result of the intervention. Category eight was defined as macro-level outcomes, compromising outcomes that expressed the impact of a programme at hospital or population level. The final category was classified as miscellaneous, covering outcomes that could not be placed in one of the other categories, but which were of importance for the patient or broader society.28

Patient and public involvement

As this paper is a review comprising an assessment of the academic literature, there was no direct patient and public engagement on the paper.

Results

Search results

The search strategy resulted in 1445 references, of which 310 were duplicates. Removing duplicates resulted in 1135 studies that were scanned on title and abstract. Full-text analysis was done on 111 articles and 59 were considered to meet the study inclusion criteria (see figure 1).

Figure 1

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram.

Baseline characteristics

Out of the 59 included studies, the majority was performed in Africa (46%) and Asia (34%). Also, the majority of the articles was published between 2012 and 2020 (64%). Out of the 59 studies, 20 studies were cost-effectiveness studies. For a complete overview, see table 1.

Table 1

General characteristics of studies included (n=59)

Characteristics	Number	Percentage of total (%)
Year
2000–2002	3	5
2003–2005	2	3
2006–2008	6	10
2009–2011	7	12
2012–2014	9	15
2015–2017	11	19
2018–2020	18	31
2021	3	5
Geography
Africa	27	46
Asia	20	34
Latin-America	8	13
Europe	3	5
Middle East	1	2
Study design
Cost-effectiveness	20	34
Quasi experimental cohort study	17	29
Randomised control trial	11	19
Quasi experimental retrospective cohort study	8	13
Retrospective case-control study	1	2
Non-randomised controlled trial	2	3
Classification of infectious diseases
Acute febrile infections	30	51
Inpatient infections (ASPs)	17
Malaria	6
Respiratory tract infection	2
Upper respiratory tract infection	2
Group of acute infectious diseases (caused by parasitic infections, bacterial infections, viral infections)	2
Postdischarge infectious disease	1
Non-acute febrile infections	22	37
HIV	17
Tuberculosis	4
HIV and tuberculosis	1
Other non-acute infections	7	12
Lymphatic filariasis	1
Schistosoma haematobium	1
Schistosoma japonicum	1
Leprosy	1
STD	1
Candidiasis	1
Soil-transmitted helminthiases and Clonorchiasis	1

ASP, antimicrobial stewardship programme; STD, sexually transmitted disease.

Interventions identified

Across the 59 studies that met the inclusion criteria, 36 unique interventions were identified (table 2). The list of interventions includes non-training and non-educational interventions that were combined with a training or educational intervention.

Table 2

Overview of interventions with number of studies reporting the respective intervention (per cent of total number of studies), categorised per healthcare value chain, per target group, per condition

Intervention	Acute febrile infections			Non-acute febrile infections		Other non-acute infections
	Patient	Physician	Non-physician professionals	Patient	Non-physician professionals	Patient	Physician
Health promotion
Media campaigns	–	–	1 (2%)	3 (5%)	–	2 (3%)	–
Improvement of basic needs (safe water, sanitation)	–	–	–	1 (2%)	–	2 (3%)	–
Primary school education	–	–	–	2 (3%)	–	–	–
Support to receive school education (non-disease related)	–	–	–	1 (2%)	–	–	–
Prevention
Free commodities supplies (soap, oral rehydration salts, mosquito nets, condoms, medication)	2 (3%)	–	1 (2%)	6 (10%)	–	–	–
Health education from health advisors	–	–	–	9 (15%)	–	2 (3%)	–
Peer-led/community-based support workers outreach and education	–	–	–	9 (15%)	–	–	–
HIV testing	–	–	–	8 (14%)	–	–	–
Prescription of preventive medication	–	–	–	3 (5%)	–	3 (5%)	–
Case finding of leprosy by dedicated team travelling from city to city	–	–	–	–	–	1 (2%)	–
Treatment
Physician instructed care support via teachers/community-based support workers	2 (3%)	–	–	1 (2%)	–	–	–
Presentation and discussion of (newly created) clinical guideline	–	13 (22%)	–	–	1 (2%)	–	1 (2%)
Training on AMR	–	15 (25%)	–	–	–	–	–
Feedback on baseline antibiotic prescription practices	–	11 (19%)	–	–	–	–	1 (2%)
Create new guideline for optimal prescription	–	10 (17%)	–	–	–	–	1 (2%)
Antimicrobial order form	–	7 (12%)	–	–	–	–	–
Review/modification of prescription by AMR team	–	5 (8%)	–	–	–	–	–
Bedside discussions among AMR expertise group	–	3 (5%)	–	–	–	–	1 (2%)
Face-to-face (individual) interactive discussions	–	4 (7%)	–	–	–	–	–
Antimicrobial susceptibility patterns shared with physicians	–	3 (5%)	–	–	–	–	–
Peer review/presentation and discussion of the guideline, and presentation of clinical scenarios	–	3 (5%)	–	–	–	–	–
Motivational interventions (fine based)	–	1 (2%)	–	–	–	–	–
Restricted use of specific drugs	–	1 (2%)	–	–	–	–	1 (2%)
Introduction of an antibiotic prescription chart	–	1 (2%)	–	–	–	–	–
Skill-based training on management of diseases	–	–	3 (5%)	1 (2%)	2 (3%)	–	–
Facilitation of community mobilisation	–	–	1 (2%)	1 (2%)	–	1 (2%)	–
Financial support (free treatment of disease, reimbursement of travel cost, care and assistance)	–	–	–	8 (14%)	–	–	–
Offering free food to reduce food insecurity and encourage clinic visits	–	–	–	2 (3%)	–	–	–
Prioritisation of patients with HIV over other patients	–	–	–	1 (2%)	–	–	–
Introduction of medication dosing table					1 (2%)
Syndromic management of STI			–	1 (2%)	–	–	–
Maintenance/postintervention care
Educational materials for caregivers, patients and communities	2 (3%)	–	1 (2%)	4 (7%)	–	2 (3%)	–
Scheduling post-discharge follow-up visits	1 (2%)	–	–	–	–	–	–
Sending postdischarge reminders for treatment adherence	–	–	–	1 (2%)	–	–	–
HIV counselling	–	–	–	7 (12%)	–	–	–
Peer support network					1 (2%)

AMR, antimicrobial resistance; HIV, human immunodeficiency virus; STI, sexually transmitted infection.

The studies in the current review described interventions targeting three different groups, including patients, physicians and non-physician professionals. The group of non-physician professionals consisted of retail shopkeepers, pharmacists and lay health workers. Most interventions were targeting patients (21/36; 58%), followed by interventions targeting physicians (13/36; 36%) and a minority targeting non-physician professionals (8/36; 22%). Some interventions were targeted at more than one group.

Among the interventions that targeted patients or caregivers, the most prevalent interventions were focused on the education of patients or caregivers by peers, community workers or health advisors. The educational goals and topics differed across the studies. Studies on HIV covered sexual and reproductive health education for adolescents and youth,29–32 and education aiming to change sexual behaviour for individuals at high risk (ie, sexually active individuals, sex workers).29 33–37 Also, studies on HIV incorporated educational interventions to prevent pregnancy-related HIV transmission38–40 and more general health education for (pregnant) women on the prevention of HIV infections.41 42 Educational interventions in studies not targeting HIV were focused on improving knowledge of the disease (ie, infections with TB, lymphatic filariasis, leprosy, malaria, soil-transmitted helminthiasis (STH)) and promoted preventive behaviour for specific groups (ie, youth, adolescents, patients, pregnant women) or across the general population.30 41 43–49

Interventions targeting the physician were mainly focused on the promotion of adequate use of antimicrobial drug therapy by physicians.50–68 In addition, physician-targeted interventions aimed to improve adequate use of antifungal therapy69 and improved management of infectious diseases.70–73

Four studies described interventions that targeted drug retail locations (eg, pharmacies, shopkeepers) that play a vital role in appropriate drug use. By improving the health skillset of people at pharmacies and drug retailers, appropriate use of antimalarials and improved syndromic management of sexually transmitted diseases (STD) was promoted.74–77 One study described an intervention that aimed to improve the knowledge and skills of lay health workers to improve TB care provided to patients and subsequently improve treatment adherence.78

Input parameters identified

A total of 42 unique input parameters were found. Categorisation of the input variables resulted in four overarching parameter types: (1) cost parameters, (2) disease-specific parameters, (3) population background characteristics and (4) intervention details (see table 3).

Table 3

Overview of input parameters

Category	Definition	Input variables	Reported in N studies (% of total; % of total within the respective category)
			Acute febrile infection	Non-acute febrile infections	Other non-acute infections
Cost	Costs related to the intervention/the programme	Programme cost:Cost of travel and accommodation for personnel; cost of buildings; cost of overhead; cost of refreshments; start-up costs; cost of training or education; programme management costs; programme development cost; programme implementation cost; recurring costs for training; personnel cost; cost of transportation of supplies; cost of equipment; cost for data capture and use;Cost of care:Routine care costs; daily cost of ICU admission; average cost of one inpatient day; cost of social mobilisation; pharmacists costs; cost of consultation; cost of lifetime treatment; cost of diagnostic tests; cost of death; cost of supplies/medication;Cost for the patient/caregiver:Travel cost; cost of time lost for caregiver; out-of-pocket costs	20 (34%; 67%)	14 (24%; 64%)	5 (8%; 71%)
Aetiology-specific characteristics	Disease-related characteristics that have impact on the intervention outcomes	ART initiation age; awareness of HIV status; bacterial resistance rates; disease transmission rates; average duration of the disease; disease prevalence	6 (10%; 20%)	7 (12%; 32%)	4 (7%; 57%)
Population background	Background information on the targeted population which could affect the outcomes of the intervention	Number of people at risk in the area; average life expectancy; average number of sex clients per month; average time span men buy sex; average time span women sell sex; proportion of individuals using condoms	–	4 (7%; 18%)	1 (2%; 14%)
Intervention details	Details of the intervention that put the intervention in a broader perspective	Number of individuals reached with the intervention; efficacy of the intervention; the proportion of the population at risk targeted by the intervention	–	5 (8%; 23%)	1 (2%; 14%)

ART, antiretroviral therapy; HIV, human immunodeficiency virus; ICU, Intensive Care Unit.

The majority of the input parameters detailed the costs of an intervention (27 unique parameters). Within the cost category, a clear distinction was present between cost related to the programme, cost for care and cost for the patient and caregiver. Great variety existed among the studies, none of the cost parameters was used across all studies.

Acute febrile infections

No consistent approach was found among studies that included cost input parameters. A large proportion of the studies only included the cost of medication, not taking any other programme or care-related costs into account.50 51 53 59 62 64 67 68 Though, there were also studies that took a more extensive approach by reporting both cost of care (eg, cost of medication, cost of consultation) and programme costs (eg, cost of personnel, cost of training and cost of programme management).55–57 60 72 75 76 79 80 Across all studies in the review, only three studies included the cost for the patient and caregiver. These studies were cost-effectiveness studies of malaria interventions performed from a societal perspective.72 76 79

Non-acute febrile infections

All non-AFI studies that reported costs as input parameters, included at least one variable on the cost of care and one variable on costs of the programme.29 30 33 34 36 37 39 41 42 45 81–84 The cost of supplies such as condoms and medication was reported most frequently.29 33 34 37 39 41 42 45 81 83 None of the studies included the costs for the patient and caregiver.

Other non-acute infections

Studies that included costs for interventions targeting non-acute infections reported costs in different ways. One study on candidiasis only included the cost of medication,69 while studies on STD, Schistosoma japonicum, STH and leprosy incorporated both costs of care and cost of the programme.43 49 74 85 None of the studies included the costs for the patient and caregiver.

Outcomes identified

A total of 81 unique outcomes were reported in 59 studies which are categorised into 9 categories (see table 4). In the section below, the five categories that were reported in most studies are reviewed in more detail.

Table 4

Overview of outcome variables

Category	Definition	Outcome variables	Reported in N studies (% of total; % of total within the respective category)
			Acute febrile infections	Non-acute febrile infections	Other non-acute infections
Final health outcomes	Quantification of the health effect of an intervention, addressing the length or quality of life	QALY; DALY; YLS; deaths averted; mortality rate; mortality increase/decrease; cured rate	11 (19%; 37%)	9 (15%; 41%)	1 (2%; 14%)
Intermediate health outcomes	Quantification of the health effects of an intervention as a change in clinical indicator that may or may not lead to final health outcomes28	Disease-specific outcomes; number of cases correctly treated; infections averted; number cases detected with disease; infection rates; recurrence rates; number of adverse drug reactions; % positive and negative tests; number of individuals receiving treatment; quality of life	19 (32%; 63%)	8 (14%; 36%)	5 (8%; 71%)
Cost outcomes	Quantification of the costs as a result of the whole programme or single intervention	Total cost; cost reduction/costs saved; cost of intervention per patient; cost per individual tested; costs per person reached; cost per 100 bed-days	18 (31%; 60%)	11 (19%; 50%)	4 (7%; 57%)
Prescription outcomes	Quantification of the impact of an intervention on prescribing practices	Antibiotic use density; DDD/100 patients; (antibiotic) prescription rate; DDD/1000 or 100 patient days; number of inappropriate prescriptions; total antibiotic days of therapy/1000 patient days; % of prescriptions containing more than one antibiotic; % of prescriptions having broad spectrum antibiotics; grams of antibiotics prescribed; number of times adjustment of antibiotic prescription done	19 (32%; 63%)	–	1 (2%; 14%)
Health economic outcomes	Outcomes reflecting the incremental cost per single unit of outcome	Cost per infection averted; cost per individual adequately treated; cost per HIV case detected; costs per averted loss-to-follow-up; cost per decrease in antibiotic prescription rate; Cost per QALY; cost per DALY averted; Cost per YLS; cost per death averted; cost per reduction in male sexual partners; cost per % increase in condom usage	6 (10%; 20%)	13 (22%; 59%)	3 (5%; 43%)
Behaviour outcomes	Outcomes that indicate the effect of the intervention on health-related behaviour of the targeted individual	Adherence rates; attrition rates (including loss-to-follow-up and mortality); number of admissions; loss-to-follow-up rate; averted loss-to-follow-up; % retention in care; completion of follow-up visits; number of referrals to secondary health clinics by GP; number of women giving birth at health facility; number of ANC visits; number of cases that did postpartum check-up; number performing exclusive breast feeding; % using family planning;	6 (10%; 20%)	10 (17%; 45%)	1 (2%; 14%)
Time-related outcomes	Quantification of the time-related component of an intervention	Time efficiency gain; time to event; duration of hospital stay; per person life-expectancy losses due to loss-to-follow-up; time till loss-to-follow-up	7 (12%; 23%)	4 (7%; 18%)	–
Macro-level outcomes	Expressing the impact of a programme/intervention at hospital or population level	% tested; medical care utilisation days; number of diagnostic tests done; ICU admissions; absolute risk ratio; number needed to treat; % receiving treatment; Bacterial resistance rates	7 (12%; 23%)	4 (7%; 18%)	1 (2%; 14%)
Miscellaneous	Intervention-specific outcomes, which are not direct measures of health but are of societal importance or of importance for the patient28	Number of times replacement drug is provided; number of male partners attending care visits; number of physicians receiving fines; number of times education provided to the patient; number of early infant diagnosis done; population knowledge of the disease; number of times combined medication provided; number of (couple) HIV testing and counselling; number of individuals with access to clean water; % increase in condom use; reduction in number of sexual partners	4 (7%; 13%)	4 (7%; 18%)	1 (2%; 14%)

ANC, antenatal care; DALY, disability-adjusted life years; DDD, defined daily doses; GP, general practitioner; HIV, human immunodeficiency virus; ICU, intensive care unit; QALY, quality-adjusted life year; YLS, years of life saved.

Final health outcomes

Out of the 59 studies, 21 studies reported final health outcomes. Final health outcomes—reported in DALYs averted, QALYs gained, years of life saved (YLS), mortality rate, cured rate and deaths averted—were found in studies across all three infectious disease categories.

Among the studies on AFI, one study on malaria reported DALYs and deaths averted, calculated based on the probability of death for a child with fever for whom treatment is first sought from a shop, with and without the intervention.75 Seven studies on inpatient infections reported mortality rates (increase/decrease) as a result of the intervention.50 54 56 58 60 67 86 One study on postdischarge infections reported final health outcomes in deaths averted, defined as hospitalised patients that survive 30 days after discharge.61

In total, six studies on HIV reported DALYs averted, calculated from the number of infections averted.29 34 36 39 41 83 Besides the studies reporting DALYs averted, there was one study on HIV reporting QALYs to quantify the impact of the prevention of mother-to-child HIV transmission.42 To estimate QALYs, the difference between the expected number of QALYs of a child living with and without HIV was calculated.42 One study on HIV reported outcomes in YLS calculated from the life years lost as a result of loss-to-follow-up from ART.81 Two studies on TB reported the final health outcomes as the number of patients cured, defined as individuals who are smear or culture negative in the last month of treatment,44 78 and another study on TB reported the outcome as the reduction in mortality rate as a result of the intervention.45

Only one study in the category of other non-acute infections reported a final health outcome. The study on leprosy reported the number of patients cured, defined as individuals completing the therapy.43

Intermediate health outcomes

Among the studies reporting on AFI, the most frequently reported intermediate health outcome was the number of patients that are correctly treated, covered in studies on inpatient infections, malaria and acute respiratory tract infections.50 51 55 56 63 66 71–73 76 76 77 The recurrence rate, also indicated as unexpected readmission rates, was reported in six studies covering inpatient infections, respiratory tract infection and postdischarge infections.54 56 58 60 67 86 Other intermediate health-outcomes reported in studies on AFI were less widely reported. These outcomes included the number of cases diagnosed with malaria,72 and the number of adverse events occurred after implementation of antimicrobial stewardship programmes (ASPs) for improved management of inpatient infections.63 64

The two most reported intermediate health outcomes in studies on HIV or TB were the number of cases diagnosed84 87 and the number of infections averted.29 34 42 Across all studies in the review, only one study reported the quality of life of the patient, which was measured using the EQ-5D with patients with TB.44 Disease-specific clinical outcomes were also found in studies on HIV and TB. Examples of disease-specific outcomes were reduced TB stigma or CD4 count slope.30 88

One study on STD reported intervention outcomes in the number of patients correctly treated.74 Two studies, on STD and candidiasis, reported the results in the number of unexpected readmissions.69 74 The number of cases diagnosed was reported in one study on leprosy43 and the increase/decrease of infections as a result of the intervention was reported in two studies covering S. japonicum and STH infections.49 85

Cost outcomes

The cost impact of an intervention was reported in an aggregate form (ie, total programme costs and total cost saved) or on a per-unit basis (eg, per person reached). The aggregated total costs of the programme/intervention34 36 39 43 49 53 57 62 65 67 68 71 75 76 79 80 82–85 and the costs saved as a result of the intervention36 42 53 54 56 57 60 60 64 67–69 were often reported across all three infectious disease categories.

Only studies on non-AFI reported the cost per unit. Three studies on HIV reported cost per person reached29 33 36 and one study on HIV indicated the cost per individual tested.33

Health economic outcomes

Only six studies in the category of AFI reported health economic outcomes, out of which four were on malaria. Studies on malaria reported health economic outcomes as the cost per case adequately treated,72 75 76 79 cost per DALY averted75 and cost per death averted.75 Cost per death averted was also reported in a study on inpatient infections.61 The cost per percentage reduction in antibiotic prescription was reported once in a study on upper respiratory tract infection.80

Health economic outcomes were most often reported in studies on non-AFI. Twelve out of the 17 studies on HIV reported on the cost-effectiveness of the intervention. Variables included were cost per infection averted,34 36 42 87 cost per QALY,42 cost per HIV case detected,84 87 cost per DALY averted,29 34 36 39 41 83 cost per averted loss-to-follow-up,30 82 cost per YLS,81 cost per reduction in male sexual partners37 and cost per % increase in condom use.37

Cost-effectiveness thresholds, which indicates the maximum amount a country or organisation is willing to pay for a unit of health-outcome, were only applied in studies on HIV. The thresholds ranged between one to five times gross domestic product per capita per DALY averted29 36 39 41 or per YLS.81 For all five studies that applied cost-effectiveness thresholds, the cost per DALY averted or cost per YLS of the interventions fell below the cost-effectiveness thresholds. Hence, these interventions were considered cost-effective compared with the standard of care.29 36 39 41 81

In the category of other non-acute infections, health economic outcomes were rarely reported. One study on S. japonica reported cost per infection averted85 and one study on STD reported the cost per case adequately treated.74

Prescription outcomes

The category of prescription outcomes included outcomes reported in studies that aimed for more appropriate use of antimicrobials and antifungals by physicians, and was predominantly found in studies on AFI and in one study on other non-AFI. The category of prescription outcomes provided insight into three main factors: (1) the overall prescription practices by physicians, (2) the quality of the prescription practices and (3) the quantitative prescription details (see table 4).

As an indicator of the overall prescription practices, three outcomes were reported: the antibiotic prescription rate (number of times antibiotics prescribed),55 57 62 65 67 69 70 80 percentage of the prescriptions containing more than one antibiotic65 and percentage of prescriptions containing broad-spectrum antibiotics.65

The quality of the prescription practices was reflected by the number of inappropriate prescriptions, defined as incorrect antimicrobial prescribed, incorrect dose prescribed, incorrect duration prescribed or incorrect decision to prescribe antimicrobials.52 62 68 69 Another outcome that indicated the quality of prescription practices was the number of times adjustment of prescription was done.50

The quantitative details of the prescription were reported in a variety of ways. Four studies reported the total DDD prescribed.64 67 68 80 The DDD is a validated method to standardise the number of doses consumed and is developed by the WHO. Nine studies reported the total DDD per 1000 patient days or 100 patients treated.51 53 54 56 59 60 67–69 One study reported the total antibiotic days of therapy per 1000 patient days, defined as the days of antibiotic therapy administered to the patients independent of the doses. The days of therapy was calculated by multiplying the number of doses received by the dosing interval (in hours) and then divided by 24 hours for each antibiotic the patient received.58 The antibiotic use density was given once, which was equal to DDD per 100 patient days, and was calculated by multiplying the DDD by 100, divided by the number of patient.66 One study reported the antibiotic prescription in total grams.68 All studies on inpatient infections that reported on antibiotic consumption reported a decrease in the total antibiotics consumed51 53 54 56 58–60 64 66–69 with some small increases on individual antibiotics.50 51 53 57 59 60 62 64 67

Discussion

The results of the current review provide insight in the wide range of programmes that aim for improved infectious disease management in LMICs. The programmes consisted of one or more interventions that span across the healthcare pathway and target different stakeholder groups including patients, physicians and non-physician professionals. The input and outcome parameters reported in the studies did not show a consistent and generalisable set of metrics used across all studies. However, by grouping the individual variables into categories, it became evident that four input categories and nine outcome categories could be considered when reporting the impact of a programme targeting infectious diseases.

Heterogeneity in outcomes is a well-known factor of influence in clinical research.89 Several initiatives have started to improve the standardisation of metrics measured and reported in clinical studies. One of these initiatives is the Core Outcome Measures in Effectiveness Trials (https://www.comet-initiative.org/) initiative, which launched in 2010 to coordinate efforts in the development of core outcome sets (COSs) across a wide range of areas of health. The definition of COS is ‘an agreed standardized collection of outcomes that should be measured and reported for a specific area of health’.90 Unfortunately, for infectious disease, the number of COS developed is limited, existing COS on infectious diseases has not been updated recently91 92 and the involvement of LMICs in the development of the COS was low.93 Therefore, we suggest that further research will continue with a critical assessment of the categories and metrics found in the current review. These efforts could function as valuable input to establish an initial COS for infectious disease management programmes in LMIC.

Reporting on final health outcomes is crucial to allow comparisons between interventions. Final health outcomes are standardised and widely used outcomes across multiple disease areas, as opposed to intermediate health outcomes that could be disease specific and thereby making it difficult to extrapolate and compare with other disease areas. The most used final health outcome in global health studies and in LMICs is the DALYs averted, which is used to define the burden of the disease.15 Also within the current review, DALYs averted were the most frequent reported final health outcome, mostly found in studies on non-AFI (eg, HIV)29 34 36 39 41 83 and only one time in a study on AFI (eg, malaria).75 Studies on AFI more often report on an increase or decrease in mortality rate. However, as opposed to DALYs, mortality rates do not quantify the impact of a disease on morbidity,94 which is why the DALY is preferred over the mortality rate. One of the potential reasons for not reporting the DALYs could be the lack of local data for estimating the DALYs, which appeared to be an important reason for researchers in LMIC to not include the DALYs averted.95 Also, infectious diseases are often self-limiting and of short duration, thereby having a small impact on the estimated DALYs per patient, but on population level could still result in a substantial disease burden.1 To encourage researchers in reporting on important outcome parameters like DALYs averted, the Guide to Economic Analysis and Research (GEAR; http://www.gear4health.com/) online resource was introduced as a reliable aid for researchers in LMICs that provide solutions for methodological difficulties.22 Although it could be a helpful resource, none of the studies in the current review mentioned or referred to the GEAR resource. Hence, further dissemination of the GEAR resource among researchers performing health-economic analyses for LMICs could be of benefit to improve standardisation across studies.

The impact of a health intervention should logically be expressed in health outcomes, but also the financial impact should be considered. Being able to compare interventions on health-related and economic outcomes, allows policy-makers to create health policy with the intervention that maximises the health impact per monetary unit spent. There are different approaches researchers could take when calculating the cost of an intervention, considering direct and indirect costs. Within the current review, most of the studies reporting the costs of an intervention only included direct costs, with substantial variations in the type of direct costs included. These methodological variations have impact on the results and make comparisons between studies less reliable. A more standardised approach for calculating costs would improve generalisability of results and thereby enhance the ability to compare outcomes between different studies. Wider implementation of existing guidelines could be an important step towards more generalisable results for studies in LMICs. For example, for health economic studies, the CHEERS provides guidance in the reporting of health economic assessments. The CHEERS guideline includes some high-level recommendations in the decision on what costs to include, depending on the perspective that is taken (eg, healthcare system, societal).25 Also, for studies on ASPs, the US guideline incorporated recommendations to include costs on programme management, salary for stewardship personnel and medication purchasing costs.96 With the US guideline for studies on ASPs and the CHEERS guideline for health economic assessments, some guidance already exists and could be more broadly applied as an initial step towards more generaliable cost outcomes.

Indicating the impact of an intervention on prescription practices has been considered as an important outcome variable. As such, standardised approaches are introduced by WHO to enable clear and concise reporting of prescription outcomes.97 Especially in the case of antimicrobial prescriptions, the dose, frequency and duration are important to assess the impact of an intervention on the consumption and the related AMR. Within the current review, the DDD was the most reported outcome in the category of prescriptions outcomes. The DDD is a standardised approach but is impacted by weight-based dosing as done for paediatrics.96 Therefore, instead, days of therapy is suggested as a more valuable parameter since it is not impacted by dose adjustments. When following the guidelines from the Infectious Diseases Society of America and the Society for Healthcare Epidemiology of America, days of therapy is the preferred option.96 In the present review, only one study reported the outcomes in days of therapy58 which could imply that the impact of weight-based dosing has been overlooked in the other studies. Moving forward, to give a more complete picture of antimicrobial prescription, researchers could consider to include the antimicrobial use expressed in days of therapy if possible.

The studies on infectious diseases that reported antimicrobial consumption in DDD or days of therapy as the main outcome measure51 53 54 56 58–60 64 66 69 did not report final health outcomes in DALY, QALY or YLS. Thereby making it challenging to compare the effect of these interventions with interventions not reporting DDDs or days of therapy. Translating antimicrobial use into a value that indicates the burden of the disease in more generalisable outcomes, such as DALYs, is challenging and comes with great uncertainty.98 Another possibility is to convert antimicrobial use to costs per antimicrobial prescribed to account for future resistance, as is done in some studies.99 100 However, these estimates also come with high uncertainty and there is a risk that the actual costs are far higher than the best estimates.101 Therefore, future research should focus on the quantification of antimicrobial use in more generalisable outcomes to better reflect the actual value of interventions that aim for appropriate antimicrobial use as part of the infectious disease management strategy.

The current literature review is limited in the following aspects: first, the variables found in this review show a high heterogeneity resulting in low generalisability. This could be a result of the wide scope of aetiologies included, in addition to the fact that the input and outcome parameters are often context specific. However, generalisability should, to a certain extent, also apply to interventions targeting different aetiologies to allow policy-makers to decide on the most cost-effective strategy. There should at least be a set of core outcomes across aetiologies that functions as the minimum of what should be included, still allowing for additional disease-specific measures to be added. Second, the results of the current review could guide researchers in the process of defining input and outcome parameters to report on for health economic research on infectious diseases but does not offer a concrete list of input and outcome parameters. Further research is needed to come to a COS for infectious diseases along with broad implementation and knowledge dissemination of currently available guidelines.

To our knowledge, the current study is the first review that provides an overview of health economic and health-outcome studies on training or education interventions for improved management of infectious diseases. Thereby, the current study offers valuable insights for future health economic assessments on programmes in which education is integral part of the intervention.

Conclusion

In conclusion, it can be said that standardisation of parameters is lacking across studies on infectious disease programmes. For input parameters, the most reported category was costs. For outcomes, studies reported most often on final health outcomes, intermediate health outcomes, cost outcomes, prescription outcomes and health economic outcomes. We recommend that further research will be performed on the definition of a COS for infectious diseases in LMICs.