Effective strategies to improve health worker performance in delivering adolescent-friendly sexual and reproductive health services

Background: Despite recognition of the important role of healthworkers in providing adolescent-friendly sexual and reproductive health services (AFSRHS), evidence on strategies for improving performance is limited. This review sought to address: (1) which interventions are used to improve healthworker performance in delivering AFSRHS? and (2) how effective are these interventions in improving AFSRHS health worker performance and client outcomes? Methods: Building on a 2015 review, a search for literature on 18 previously identified programs was conducted to identify updated literature and data relevant to this review. Data was systematically extracted and analyzed. Results: Due to the parent review’s eligibility criteria, all programs included health worker training. Otherwise, supervisionwas themost frequently reported interventionused (n=10). Components and methods related to quality of trainings and supervision varied considerably in program reports. Nearly half of programs described employing processes to ensure availability of basic medicines and supplies (n=7). Other interventions (policies, standards, and job descriptions [n=5]; refresher trainings [n=5]; job aids or other reference material [n=3]) were less commonly reported to have been employed. No discernible patterns emerged in the relationship between interventions and outcomes of interest. Conclusions: Multi-faceted complementary strategies are recommended to improve health worker performance to deliver AFSRHS; however, this was uncommonly reported in the programs that we reviewed. Effectiveness and costeffectiveness evaluations of interventions and intervention packages are needed to guide efficient use of limited resources to enhance health worker capacity to deliver AFSRHS. In the interim, programs should be developed and implemented based on available existing evidence on improving health worker performance within and outside adolescent health. Implications and contribution: This review is the first to examine the interventions commonly used to improve health worker performance in delivering AFSRHS. The findings indicate a need for additional effectiveness and cost-effectiveness evaluations of such interventions. In the meantime, existing evidence on improving health worker performance within and outside adolescent health must be integrated more thoughtfully into program planning and implementation.


Introduction
Adolescent-friendly health services (AFHS) are defined as those that are accessible, acceptable, equitable, appropriate, and effective [1]. Adolescent-friendly health workers are a crucial component of AFHS, and require specific competencies (defined as the knowledge and skills to carry out their required roles and responsibilities, as well as judgment on when to do so) and attitudes to provide evidence-based, nonjudgmental and non-discriminatory care, especially with regards to sexual and reproductive health (SRH). Specifically, they must be able to consider the evolving cognitive, emotional, and social capacities of adolescents, and understand and evaluate each adolescent as an individual at a specific point of development within a unique social context. Furthermore, health workers should have a clear understanding of their roles and responsibilities, be supported to carry them out, and be held accountable for doing so.
Therefore, this review aims to examine interventions (or packages of interventions) used to improve health worker performance in delivering AFSRHS. To achieve this aim, this review sought to answer the following research questions: (1) which interventions (or packages of interventions) are used to improve health worker performance in delivering AFSRHS? and (2) how effective are these interventions (or packages of interventions) in improving health worker performance for AFSRHS delivery, client satisfaction, and health outcomes?

Methods
In 2015, our group conducted a review of effective interventions to improve AFSRHS, which identified four important components in improving uptake of SRH services or commodities and related clinical outcomes among adolescents: (1) generating demand among adolescents, (2) sensitizing communities to improve acceptability of SRH services for adolescents, (3) modifying facilities to make them more adolescent-friendly (e.g., expanded out-of-school hours, structural elements to maintain privacy and/or confidentiality), and (4) providing health workers with training to improve their competencies, attitudes, and practices in providing AFSRHS [30]. In this current review, we focused on the last component and closely re-examined the 18 programs in the original review to assess the types and characteristics of trainings that were provided, as well as other interventions used to improve health worker performance, and their effects on outcomes of interest.
Workforce performance is influenced by many factors interacting in a complex fashion at multiple levelsfrom national policies and regulatory frameworks across sectors (e.g., finance, health, education) to individual health worker characteristics and circumstances [31]. In this review we focus on workplace level interventions that might influence health worker performance.
Dieleman and Harnmeijer noted that there are an "abundance of theories" explaining health worker behaviors and practices, but evidence of the extent to which these theories actually predict behaviors and practices is limited and studies on health worker performance in LMICs are not often based on an individual theory or specific combination [31]. Therefore, we did not limit this review to programs utilizing a specific theory of change and included programs regardless of whether any theoretical approach was used. Our conceptual framework was that interventions that increase health worker competencies (e.g., knowledge and understanding of how to provide evidence-based care, skills in providing such care) and/or attitudes (e.g., nonjudgmental attitudes towards adolescent sexuality) could lead to changes in performance 1 (e.g., adherence to evidence-based care guidelines, empathetic communication with clients), which could in turn result in improved client satisfaction, increased health services utilization, and/or improved health outcomesthe latter mediated through increased service utilization and/or improved quality of delivered care.

Search strategy
Information was drawn from existing literature identified in the parent study [30]. In addition to the publications on the programs examined in our 2015 review, PubMed, Google and Google Scholar were searched for more recent literature on these programs, which may have been subsequently published [30]. Specifically, the strategy included search terms for names of programs, authors, and locations. 1 The World Health Organization defines health worker performance as a workforce that is available, competent, productive and responsive [65]. In this review, we included the latter components, but did not assess availability (e.g., retention, lack of absenteeism). Competency includes provision of care and advice that is adherent to guidelines. Productivity includes number of visits, services, or commodities provided. Responsiveness includes client perceptions of quality of care.
The inclusion and exclusion criteria, as was used in our 2015 review, followed a World Health Organization framework typology for systematic reviews [32].

Data extraction
The health worker performance interventions used by the programs were assessed and classified as: health worker trainings; refresher trainings; job aids/desk references/ other written or electronic reference materials; processes to ensure that basic medicines/supplies/equipment are in place; policies/standards/job descriptions; and supervision, including trainings for supervisors. These categories were generated based on groups in existing literature, and were refined based on the descriptions of the interventions used by the programs.

Outcome measures
The outcome measures extracted in this review included several that were included in the original review (and updated for this review), including patient satisfaction, health service and commodity utilization, and biologic outcomes. We also included several new outcome indicators more proximally related to the interventions under study in this paper, including health worker competencies, attitudes, and practices.

Findings
Our original review included 18 programs; one program that was included in the parent review was excluded from this review due to lack of accessibility of the original article [33]. The search for updated information on the programs identified a scale-up follow-on of one of the original studies [34] that differed in approach from the initial project [35]; hence we assessed this as an independent program. Table 1 contains descriptions of the 18 programs included in this review and abstracted evidence on the interventions and outcomes of interest.

Description of study designs
Study designs included: case-control pre-and postintervention assessments (n=10), case-control post-intervention assessments (n=4), pre-and post-intervention assessments (n=2), post-intervention assessments (n=1), and longitudinal assessments of health service utilization data (n=6). With regards to methods for assessing health worker performance, adolescent mystery client visits were the most common method reported, followed by in-depth interviews and focus group discussions.

Overview
Due to the content area of the 2015 parent review's eligibility criteria, all of the programs included health worker trainings on AFSRHS [30]. Among the other health worker performance interventions, the most frequently reported was supervision (n=10), followed by processes to ensure that basic medicines/supplies/equipment are in place (n=7), policies/standards/job descriptions (n=5), refresher trainings (n=5), training for supervisors (n=5), and job aid/ desk reference/other written or electronic reference material (n=3). The majority (n=14) of the studies reported implementing two or fewer health worker performance interventions beyond trainings, including four that did not report any interventions beyond trainings.

Trainings
Most descriptions of the trainings were limited and did not include specific information of interest to this review. Training content was primarily described as ASRH curriculum (n=12), communication techniques (n=4), and counseling techniques (n=4), while the remainder were not specified (n=3). Only four programs described their trainings as interactive and/or participatory.
Training of trainers (TOT) strategies were reported in seven programs. Of these, one described using a training consultant and another reported using a training group (consisting of Ministry of Health (MOH) staff, program technical officers, and training consultants) to train trainers, while five programs did not specify the training personnel type. The remaining programs delivered trainings directly to health workers by MOH staff (n=1), program staff (n=2), or unspecified personnel (n=8). Training manuals were mentioned in six of the 18 programs, while training manuals for TOTs were mentioned in five of the seven programs using this approach.
Training location was noted for 10 programs; only one described conducting the trainings at service providers' work sites, while the remainder were noted as off-site. Training durations were usually not indicated (n=10), but when specified ranged from 3 to 5 days (n=2) or 6 to 10 days (n=4). Only one program specified training group size, with an average of 27 trainees per session, and no program  year olds. Many baseline reported behaviors regarding SRH commodity use were better among males prior to exposure to intervention. SRH commodity use was significantly higher among intervention exposed compared to control females; no statistically significant difference was found among males. Measure of change from baseline comparing exposed to unexposed was not provided, nor was data from individuals in intervention areas reporting no exposure to the intervention.
There was a steady increase in clinic attendance over five quarters, but then a decrease in the final quarter. Significance testing was not provided. F: Not reported Evaluation data was collected from a subset of program facilities. For example, no mystery client visits of GHS public clinics were conducted. There was some contamination of control communities due to mass media and peer education intervention components.
Denno et al.: Health worker performance for ASRH services pre-and post-test scores were  and %, respectively. B: Not reported C & D: Exit interview and mystery client visit results were largely positive; for example, % of exit interviewees reported satisfaction with their treatment. E: Reported use of condoms and contraceptives were assessed among - year olds and were significantly higher among females in intervention areas who reported being exposed to the intervention compared to controls at endline. Many condom and contraceptive use outcomes were also significantly higher among exposed Evaluation data was collected from a subset of program facilities. There was some contamination of control communities due to mass media and peer education intervention components.  provider training evaluation, which noted improved attitudes toward serving adolescents. C & D: Mystery client visit results were largely positive; for example, % reported being satisfied with the services overall. E: Reported use of condoms and contraceptives were assessed among - year olds and rates were significantly higher among females in intervention areas who reported being exposed to the intervention compared to controls at endline; no statistically significant difference was found among males. Change from baseline not reported, nor was data from individuals in intervention areas reporting no exposure to the intervention. There was a steady increase in clinic attendance over five quarters, followed by a decrease in the final quarter. Significance testing was not provided and data was incomplete across all quarters and all facilities. F: Not reported analysis of data, and critical staff turnover at the end of project. Baseline data was retroactively scored from qualitative data and the scoring of the initial assessment of facilities was not done at the time of the baseline assessments. There was some contamination of control communities due to mass media and peer education intervention components. On-the-job training covered being welcoming, maintaining non-judgmental attitudes, offering minimal waiting time, privacy, confidentiality and affordable services. Quasi-experimental with longitudinal assessments comparing two intervention sites and one control site. I : Reproductive health education to out-ofschool adolescents linked with AFHS.  service providers trained. I : As per I  + schoolbased education and linkage with health facility.  service providers trained. C: Control Before and after householdbased survey of adolescents intended to represent sample of adolescents in the sites.  adolescents (∼ from each intervention and control sites before and after intervention) interviewed. Also facility service utilization data.
A: The project's health provider training was considered to be refresher training because the providers had already received AFHS training. Therefore, the project training served to reinforce prior training. B: A flip chart on ASRH was developed and distributed to health service providers (six each in intervention clinics) as well as to teachers (I ) and adolescent facilitators (I  and I ). C: Not reported D: Not reported E: Not explicitly described in regard to health providers, but it appears that clinic managers were involved in monitoring and supervising the school-based component.
A: Not reported B: Not reported C: Not reported D: There was no clear pattern of differences between intervention and control sites in terms of attitudes toward health facility contraceptive and STI services among adolescents (regardless of whether they had attended the health facilities); in some instances, control participants' attitudes toward services were better than intervention participants. Authors only presented endline attitudes; changes over time were not reported. E: Condom use among unmarried adolescent males did not improve; data not presented among female participants. Utilization of SRH services increased more in the intervention clinics (especially in I ) compared to the control clinic but statistical testing was not provided. F: Not reported The study's primary aim was to assess out-ofschool and in-school reproductive health education. Since health worker training was included in both intervention groups along with the education, it is not possible to determine what contribution the health worker performance interventions had on the outcomes compared to the education and community sensitization components.  A: Not reported B: Not reported C: Mystery client visit analysis was not quantitative, but generally showed that intervention health workers tended to engage in adolescent-friendly practices more often than control health workers, including No control sites. Significance testing not provided. The authors speculate that the initial increase in pregnancy rates was due to "girls … having unprotected sex with older people during the school long vacation. As a response, the program" introduced student education on intergenerational sex as a health threat. Average number of pregnancies per female student population not   newsletters, and job aids, including "flip charts, checklists, posters, and the patient register, [to] serve as tools for assuring client informed choice and following protocols and procedures." C: PSI ensures starter stocks of products (e.g., the pill, injectables, condoms, STI treatment kits, IUD insertion kits, and disinfection equipment) for free to the clinics. As a condition of membership, the clinics must ensure sufficient stock of FP methods, condoms, and condom demonstration models and meet minimum equipment stipulations. D: "All franchise clinics meet minimum 'youthfriendly' standards relating to facilities, administrative systems, and staff and providers -latter characteristics included: Trained to serve youth, respect for youth, maintain confidentiality and privacy, allow enough time to receive and engage in exchange with clients, and serves young men as well as young women." E: Monthly (at a minimum) regional supervisor adolescent visits in year one to over , in year ). Data regarding visits per clinic were not reported. F: Not reported Moderate evidence of increased reported use of condoms and contraceptives as a result of the intervention. A follow-up was also conducted  months after the program ended. The longer-term effect at the community level (i.e., including those who were not exposed to the intervention (especially younger adolescents who "aged into" the assessment age range)) on contraceptive use did not persist. However, the increased contraceptive use effect was retained among the sub-set of adolescents who were exposed to the intervention (albeit with a diminished but still Statistical analysis did not take into account clustering. Number of health workers trained and number of clinics in control and intervention sites not reported. Ability to evaluate long term impact was hampered by the non-program related SRH IEC activities that were introduced in both the intervention and control communities.     reported on the composition of trainee groups (e.g., by health worker cadre). Only one program indicated that its curriculum was also integrated into pre-service provider training.

Refresher trainings
The five programs that reported refresher trainings indicated a variety of approaches. One program based their one year training-to-refresher training interval based on process evaluations, including supervisory assessments. One program provided refresher trainings after two years but did not specify the rationale for the interval. A different program considered their initial trainings to be a refresher training, as the health workers had previously received AFHS training. Another program provided various opportunities for health workers to access continuing education as an incentive mechanism for complying with program requirements. Finally, one program provided refresher trainings as part of the ongoing supervision technical support. Meanwhile, qualitative information from the reports reiterated the need for refresher trainings. For example, one report noted that while half of "nurses had implemented. E: Not reported more likely to feel that provider practices were adolescent-friendly; clinic staff surveys asking about clinic providers generally (i.e., not specific to the interviewee) corroborated this. D: There were no statistically significant differences in health provider characteristics that adolescents visiting intervention facilities liked best and least about facility staff. Overall, adolescent clients were three-fold more likely to report satisfaction with services at intervention compared to control facilities. E: Weak evidence of increase in service utilization as a result of the intervention. F: Not reported strongly associated with client satisfaction.
received some type of adolescent-friendly health service training … all indicated that they needed more training in working with adolescents" [61]. Another program noted that ongoing training needs assessments could be used to identify specific areas of weakness, which could then be emphasized in periodic refresher trainings [42].

Job aids/reference materials
Of the three programs that reported providing job aids/ reference materials, one provided flipcharts to service providers. Another provided handouts to participants during refresher trainings. The other provided an assortment of materials, including flip charts, checklists, and posters.
Processes to ensure that basic medicines/ supplies/equipment are in place Of the seven programs that reported processes to ensure that basic medicines/supplies/equipment are in place, three used facility assessments to determine whether a sufficient supply of a minimum package of services were available. With regard to the specific medicines/supplies assessed, three programs focused on essential SRH equipment and supplies and one focused on STI medicines. Another program provided clinics with an initial starter kit of necessary supplies and equipment and subsequently used assessments of basic medicines, supplies, equipment, and services as a condition for membership in the program.

Policies/standards/job descriptions
Of the five programs that reported providing policies/ standards/job descriptions, one developed a training manual for health workers, which included international covenants, national policies, standards of practice, and institutional procedures. The other four programs mentioned standards for the provision of AFHS; one of these specified that the standards were developed by the MOH and one specified that the standards were developed by nurses and then independently assessed.

Supervision, including trainings for supervisors
Ten programs reported supervisory activities using a range of approaches. All reported the frequency of supervisory visits, which occurred on a quarterly (n=4), monthly (n=4), and "regular" (n=2) basis. The supervisors themselves were described as program staff (n=5), trained supervisors (n=5), MOH personnel (n=2), regional/district supervisory teams (n=2), and/or health clinic staff/managers (n=2). The focus of the supervisory visits in five of the 10 programs was to review implementation of planned activities and identify challenges and corresponding solutions. Additionally, one of these programs provided technical assistance for the provision of counseling by trained psychologists during the supervision visits. The focus of the visits in two other programs was to assess clinic accessibility in another program, and to monitor national standards and document satisfaction of clients. Two programs did not specify the focus of the visits.

Associations between strategies and outcomes of interest
To attempt to link health worker performance interventions to outcomes (health worker competencies, attitudes, and practices; patient satisfaction; health service and commodity utilization; and biological outcomes), we used a number of analytic strategies. These strategies included frequency tables, heat maps by positive and negative outcomes, and cross-tabulation. However, no discernible patterns emerged in terms of the number or types of interventions and their relationship to outcomes of interest.

Discussion
No discernible patterns between health worker performance interventions and outcomes could be extracted through this review. The wide variety of measurement tools and techniques, time frames, and outcome indicators used in the reports prohibits direct comparisons between interventions and outcomes. Additionally, the differences in type and strength of the study designs, and the corresponding strength of the evidence, varied tremendously across programs. However, several important themes can be drawn from this analysis when considered alongside evidence from the literature on the efficacy and effectiveness of specific health worker performance interventions.

Trainings
Training of health workers is one of the most commonly used interventions to improve health worker performance [2]. It is also one of the most effective interventions for doing so, as determined by the health-care provider performance review (HCPPR), a recent systematic review of strategies to improve health-care provider performance in LMICs [2]. However, evidence clearly suggests that certain types of trainings do not substantially or sustainably improve the competencies or attitudes of health workers [2,65]. Careful consideration must be given to the training methodologies, preparation of trainers, location, duration, and group size and composition.
With regard to training methodologies, there is good evidence that participatory and interactive methodologies are much more effective in improving health worker competencies [2,65]. It is thus highly concerning that only four of the 18 programs explicitly described participatory and/or interactive training methodologies. Although it is possible that more of the programs used participatory and/or interactive training methodologies without reporting them, this absence of information suggests an urgent need to reassess the factors involved in health worker performance interventions that we consider important enough to report on in studies. For example, without active learning opportunities to practice knowledge and skills in response to real-life problems, health workers are less likely to develop confidence in their abilities and transfer the knowledge and skills into practice [66,67].
Preparation of trainers has important implications for the quality of trainings [68]. The majority of programs that we reviewed did not provide any information as to how trainers were trained to carry out their duties. Further, because one-third of the programs did not specify their training personnel, let along the preparation that was provided to them, there are major barriers to assessing the quality of the training that was delivered. However, the fact that more than half of the programs described used training manuals of some sort may have encouraging implications for the quality and consistency of trainings.
With regard to training location, evidence suggests that while off-site trainings may be more convenient for programs, on-site trainings may be more effective in building health worker competencies [65,69,70]. It is thus logical, but perhaps concerning, that the vast majority of the programs that reported on this aspect of trainings held them in off-site locations.
With regard to training duration, evidence supports longer durations for multi-topic trainings with interactive methodologies; however, there are practical implications. Extended durations risk drop-out and place extra burden on the health centers where the trainees work [70]. These trade-offs were reflected by the fact that that durations of trainings among the programs that specified this information ranged from three to 10 days. Evidence also suggests smaller groups allow for more participatory engagement [65,66,69]. Meanwhile, there is a need for more research regarding when mixed gender and mixed cadre trainings should or should not be used. It is thus disappointing that only one of the 18 programs reported training group sizes and no programs reported information on group composition.
Taken together, these findings are discouraging both in terms of the design of training interventions and the lack of information included in program reports on important features of trainings. However, these findings are also not surprising; the pressure to train large numbers of health workers within short time periods with limited financial and human resources often requires concessions, and publication standards often limit the information that can be presented in journal articles.

Refresher trainings
Lack of opportunities to practice newly acquired knowledge and skills will result in waning of these competencies [71][72][73]. With regard to AFSRHS, this may be especially relevant in clinical environments with limited adolescent patient volumes. As such, regular refresher trainings are critical for ensuring retention of skills and knowledge [73,74]. It was thus disappointing that only five of the 18 programs reviewed reported refresher trainings, despite the recurring theme in many of the reports regarding the need for refresher trainings.

Job aids/reference materials
It is well-established that disseminating job aids/reference materials as a solo intervention is generally ineffective at improving health worker performance [2]. However, combined with other interventions (such as trainings, peer group discussions, or supportive supervision) these materials can support health worker performanceif the information is accessible and relevant and if health workers are confident that the information is accurate [65]. It is worth noting that only three of the 18 programs in this review reported this potentially useful to ensure that basic medicines/supplies/equipment are in place Lack of medicines, supplies, equipment, infrastructure, and essential amenities such as water supply, sanitation and electricity are common problems that impede the ability of health workers to provide health services, including AFSRHS [31]. These health systems factors are 7often overlooked during program design and implementation. Our findings echoed this problem, as less than half of programs reported processes to ensure that basic medicines/supplies/equipment/services were in place.

Policies/standards/job descriptions
Policies and standards communicate approved norms to health workers, and job descriptions clarify their roles and responsibilities. When they are communicated and applied, these types of guidance can help health workers be clear about what they are and are not responsible for, and may help improve job satisfaction [65,74]. As only five of the 18 programs reported developing and/or providing policies, standards, and job descriptions, this appears to be an under-utilized intervention.

Supervision
Supervision can be a powerful tool to support health workers to carry out their roles and responsibilities and hold them accountable for doing so. While supervision on its own has been shown to have only moderate effects on performance, evidence suggests that alongside additional interventions, supervision can improve performance, increase motivation, and enhance job satisfaction [2]. The increasing attention to supervision in the health sector was reflected in the findings of this review; supervision was the most frequently reported health worker performance intervention after training.
Supervision style is critically important to its success. If supervision is supportive, consistent, educational, and specific, it can have a large positive impact on health worker motivation, job satisfaction, and performance [64]. On the other hand, when supervision is purely administrative, or worse yet fault-finding or punitive, it can result in more negative than positive effects [64]. Unfortunately, supervisory systems are often of poor quality, and supervisors commonly lack the skills, tools, and time with health workers required for a supportive style of supervision. Furthermore, the focus of supervision may be less on health worker performance and health worker support and motivation and more on behaviors such as obedience, punctuality, and respectfulness [75]. While some reports included in this review mentioned the frequency and focus of supervisory visits, information was insufficient to assess the style and quality of supervision.
Just as with trainers, the preparation of supervisors is critical for delivering quality supportive supervision. Supervisors are often selected based on seniority or strong performance in a previous role (i.e., health worker, administrator), which may or may not translate into strong performance as a supervisor [31]. As such, supervisors themselves also require training and support. This was reiterated in the findings of this review; for example, one program noted that health workers identified lack of support from and competing priorities of supervisors as a primary barrier to providing AFSRHS [36].

Group problem solving and collaborative learning
Although not addressed in this review, two health worker performance interventions that have potential to support AFSRHS are group problem solving and collaborative learning. While group problem solving can be used as a participatory methodology within trainings, it can also be used as a separate intervention to improve health worker performance on an ongoing basis. Delivering AFSRHS requires health workers to respond to nuanced and complex social, ethical, and legal situationssuch as an adolescent who becomes pregnant but does not want her parents to know, or one who presents with abortion-related complications in a context where abortion is illegal and sexual activity at her age (even when consensual) is considered statutory rapewith sensitivity and empathy. Health workers must apply specific competencies in a specific manner and, where relevant, must recognize and put aside their own biases and beliefs to serve the best interests of their patients. Group problem solving and collaborative learning, whereby peers engage in bottom-up problem identification on an ongoing basis, share their experiences and learn from each other, can promote contextualized problem-solving and creation of communities of practice whereby quality AFSRHS becomes normalized and expected within the local health system's environment. The HCPPR identified group problem solving as one of the most effective interventions for improving health worker performance [2].

Multi-intervention approaches
Despite being one of the most commonly used interventions, there is good evidence that the effect of training alone on health worker performance is minimal for lay health workers and only moderate for professional health workers [2]. Furthermore, while trainings may improve health worker competencies or attitudes in the short-term (e.g., as evidenced in post-training assessments), they do not guarantee retention or translation of these new competencies or attitudes into practice in the workplace [31]. Meanwhile, evidence also suggests that multi-intervention approaches addressing multiple determinants or levers of performance, especially at various levels of the health system, may be more effective at improving health worker performance than single interventions, including training [2,31]. However, the majority of the studies in this review reportedly used two or fewer health worker performance interventions beyond training, including four that did not report any interventions beyond training. While it is important to consider using multi-intervention approaches to take advantage of their synergies, decisionmakers should be cautioned against the assumption that increasing the number of interventions will automatically increase the effectiveness of the approach in improving health worker performance [2]. Especially where resources are limited, approaches that contain fewer interventions, but which strategically select the interventions based on evidence and context and deliver them with quality and fidelity, may have greater potential to achieve the desired result.

Limitations
This review is limited in that it is not a systematic review; instead, we re-examined the programs assessed in the parent 2015 review [30]. Additionally, we were unable to identify associations between the types and combinations of health worker performance interventions and the outcomes of interest, due to a number of challenges described previously. Attempts to determine whether relationships between improved knowledge and attitudes among health workers and patient-level outcomes (e.g., patient satisfaction, health service and commodity utilization, and biologic outcomes) are mediated through improved health worker performance were similarly constrained for these reasons. Furthermore, the data included is limited to the information reported in the studies. It is possible that programs may have used interventions or achieved outcomes that were not reported, especially in journal articles which may have been constrained by publication standards which limit the information that can be presented. This may have challenged our ability to identify and describe the strategies that were employed and associate them with outcomes of interest. The lack of reported information on costs of interventions similarly hindered our ability to examine intervention cost-effectiveness.
Additionally, this review is restricted to workplace level interventions and does not address wider health systems and community factors that either help or hinder health workers' abilities and/or motivation to apply their competencies in practice. For example, staff shortages and turnover can impact health workers' workloads and the time available for clinical interactions with individual patients [31]. Low pay, lack of opportunity for career progression, and limited recognition has consequences for health worker motivation and retention [76]. Likewise, if there is lack of recognition of, or opposition to, the need for adolescent SRH services by the community, health workers may be unwilling to risk their reputation, or even their personal security, by providing AFSRHS [77]. Health systems and community factors are important considerations, especially given that job satisfaction and motivation can be adversely affected if health workers have the required competencies to provide AFSRHS and are obliged to do so, but are unable to because of issues outside their control. Lastly, we did not examine accountability mechanisms (i.e., that hold health workers responsible for their performance) outside of supervision. We recognize, though, that regulation through professional associations or civil society/community groups or by means of regulatory frameworks may also influence health worker performance [78].
Despite these limitations, to our knowledge this paper is the first to review the health worker performance interventions commonly used to support AFSRHS. AFSRHS require competent, confident and empathic health workers who can deliver quality services. Thus, efforts to improve health worker performance in AFSRHS need to be maximized. The findings of this review indicate that there is a critical need for ongoing and systematic evaluations of interventions to improve health worker performance to build an evidence-base for effectiveand cost-effectivestrategies in diverse settings. In the meantime, the evidence we do have on improving health worker performance within and outside adolescent health must be integrated more thoughtfully and concertedly into program planning and decision-making. Author contributions: VC conceived the review. DD prepared the evidence tables. MP and VC drafted the paper. DD, MP, and VC reviewed, revised, and finalized the paper. Competing interests: Authors state no conflict of interest.