Human resource management strategies and performance of donor funded HIV programmes in Kenya, 2025: a cross-sectional mixed-methods study

¹Chandaria School of Business, United States International University - Africa, Nairobi, Kenya

^&Corresponding author
Mercy Njura Njeru, Chandaria School of Business, United States International University - Africa, Nairobi, Kenya

Introduction    

Human immunodeficiency virus (HIV) remains a major global public health challenge despite significant advances in prevention, diagnosis, and treatment. By the end of 2023, approximately 39.9 million people were living with HIV worldwide, with sub-Saharan Africa continuing to bear the greatest burden of infection and treatment needs [1]. International donor initiatives, including the President's Emergency Plan for AIDS Relief (PEPFAR) and the Global Fund, have transformed HIV service delivery through expanded access to antiretroviral therapy, strengthened laboratory systems, and increased investments in prevention and community-based interventions [2,3]. These investments have contributed substantially to reductions in HIV-related morbidity and mortality while strengthening health systems in many low- and middle-income countries. However, growing emphasis on donor transition, domestic financing, and institutional sustainability has raised concerns regarding the long-term performance of donor-supported HIV programmes [4]. Recent disruptions to international HIV financing have made these concerns immediate, with policy analyses and modelling studies warning of substantial service delivery and workforce consequences where donor support contracts without a planned transition [5,6].

Human resource management (HRM) is central to the delivery of donor funded HIV programmes because HIV service delivery depends on multidisciplinary teams comprising clinicians, nurses, counsellors, laboratory personnel, pharmacists, programme coordinators, monitoring and evaluation officers, and community health workers [7,8]. In this paper, HRM is used throughout to denote the organizational systems and practices through which programme and health facility staff are recruited, compensated, developed, integrated and deployed; no distinction is intended between this term and expressions such as workforce systems or workforce management, and a single term is used for consistency. Effective HRM practices, including employee recruitment, compensation, training and development, employee integration, and employee utilization, contribute to workforce productivity, accountability, service quality, and organizational effectiveness [9]. Conversely, weak HRM systems are associated with staff burnout, absenteeism, high turnover rates, reduced morale, and disruptions in service delivery [8]. Across sub-Saharan Africa, shortages of skilled healthcare workers, inadequate workforce planning, weak supervision systems, and limited retention mechanisms continue to constrain the effectiveness of HIV programmes and broader health service delivery systems [2,10].

Global health workforce projections indicate persistent shortages and inequitable distribution of health personnel, with the largest relative deficits projected for low- and middle-income countries and for the WHO African Region, posing significant challenges to universal health coverage and to the sustainability of health programmes [11]. Regional assessments document low health worker densities relative to service needs, uneven distribution between urban and rural areas, and limited institutional capacity for workforce planning, employment and retention across the African Region, including Kenya [10,12]. Human resource management (HRM) theory and empirical evidence suggest that organizations with well-developed HRM systems are more likely to achieve positive organizational outcomes through improved employee motivation, commitment, productivity, and performance [13]. Training effectiveness studies further indicate that employee motivation and supervisory support are associated with the transfer of knowledge and skills into workplace performance [14]. Similarly, contemporary HRM literature emphasizes the strategic role of these systems in enhancing organizational effectiveness [15]. Within the health sector, workforce interventions such as supportive supervision, performance feedback, training, mentoring, and retention strategies have been associated with improvements in service quality, healthcare provider performance, and programme outcomes [16].

Evidence from HIV programmes further indicates that strong HRM systems support patient retention, service continuity, programme resilience, and the ability of organizations to respond to operational disruptions and public health emergencies [17]. In Kenya, donor-funded HIV programmes remain central to the national HIV response and continue to support treatment services, prevention interventions, diagnostics, and workforce development initiatives [18]. Nevertheless, health system challenges associated with decentralization, workforce management, financing constraints, and organizational capacity continue to affect programme implementation and sustainability [19,20]. Although donor-supported HIV programmes have achieved notable gains in service coverage and health outcomes, concerns persist regarding the sustainability of workforce investments and the contribution of HRM practices to long-term programme performance [21,22].

The relationship between HRM and organizational performance has been extensively theorized within the resource-based view and strategic HRM perspectives, which propose that human resources constitute a valuable organizational asset capable of generating sustained competitive advantage when they are rare, difficult to imitate, and supported by organizational systems that exploit them [23,24]. Much of the supporting empirical literature is observational, and associations reported in that literature should not be read as established causal effects. In the health sector, evidence suggests that workforce performance is related not only to staffing levels but also to the quality of recruitment systems, compensation structures, professional development opportunities, employee integration mechanisms, and workforce utilization practices [25,26]. At the same time, growing concerns regarding the sustainability of HIV financing have intensified the need for efficient HRM systems capable of supporting programme performance within constrained resource environments [27].

Despite growing recognition of the importance of HRM for programme effectiveness, existing research in Kenya has largely focused on HIV financing, epidemiological outcomes, donor transition, and broader health system strengthening initiatives. Limited empirical evidence exists on how far specific HRM strategies are institutionalized within donor-funded HIV programmes, and on how they relate to programme performance. Furthermore, studies examining workforce management have often concentrated on isolated practices rather than examining employee recruitment, employee compensation, training and development, employee integration, and employee utilization together.

Study objectives: the specific objectives of this study were: (i) to describe the extent of implementation of HRM strategies, comprising employee recruitment, employee compensation, training and development, employee integration and employee utilization, among donor funded HIV programmes in Kenya; (ii) to describe programme performance among donor funded HIV programmes in Kenya based on senior programme managers' ratings of service delivery effectiveness, achievement of programme targets, organizational efficiency and continuity of HIV service provision; (iii) to determine the direction, strength and statistical significance of the association between HRM strategies and programme performance using Pearson correlation and unadjusted linear regression; and (iv) to explore how senior programme managers and other key stakeholders explain the implementation of HRM practices and constraints affecting programme implementation.

Research questions: the study addressed three research questions: (1) What is the extent and pattern of implementation of HRM strategies among donor funded HIV programmes in Kenya? (2) How do senior programme managers rate the performance of the donor funded HIV programmes in which they work? (3) What is the direction, strength and statistical significance of the association between HRM strategies and programme performance, and how do qualitative accounts from key informants explain, complement or diverge from the quantitative findings? Consistent with the cross-sectional observational design, the study questions focused on description and association rather than causal effects.

Methods     

Study design: this study employed a convergent mixed methods design [28]. The quantitative component used a cross-sectional, non-experimental descriptive correlational survey to describe the implementation of HRM strategies and determine their statistical association with programme performance. The qualitative component comprised semi-structured key informant interviews to provide contextual accounts of HRM practices, implementation experiences, and constraints affecting donor-funded HIV programmes. The convergent design was appropriate because the study sought to examine both the extent of HRM implementation and its association with programme performance while also exploring how managers and other key stakeholders understood and experienced the HRM practices and constraints underlying programme implementation. Quantitative and qualitative data were collected concurrently during 2025 and analysed independently before being integrated at the interpretation and reporting stages. Reporting of the quantitative component followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement [29], and reporting of the qualitative component was guided by the Consolidated Criteria for Reporting Qualitative Research (COREQ) checklist [30].

Study setting: the study was conducted in 2025 among donor-funded HIV programmes implemented across 40 of Kenya's 47 counties. The programmes were supported through implementing partner organizations, county HIV coordination structures, and donor-supported health facilities, reflecting the decentralized organization of HIV service delivery in Kenya. The study setting encompassed a multisectoral programme environment involving national and county governments, development partners and implementing organizations, with human resource management responsibilities distributed across these structures. The 40 counties represented the geographical coverage of the donor-supported HIV programme structures included in the study sampling frame, while the remaining seven counties were outside the study frame.

Participants: the quantitative target population comprised 4,488 senior HIV programme managers drawn from three strata: implementing partner organizations, county HIV coordination structures, and donor-supported health facilities across 40 counties in Kenya. Eligible managers were those with direct responsibility for programme planning, human resource management, resource allocation, implementation, monitoring, or reporting within donor-funded HIV programmes. For the qualitative component, 20 key informants were purposively selected from programme stakeholders with in-depth knowledge of resource management and programme implementation. They included senior programme managers, county HIV coordinators and implementing partner representatives, selected on the basis of their direct involvement in programme planning, budgeting, resource allocation, monitoring and reporting and their experience within donor-funded HIV programmes.

Study size: the minimum sample size was calculated using Yamane's formula for a finite population [31]

Where n is the required sample size, N is the total target population, and e is the desired level of precision (margin of error). Based on a target population of 4,488 senior HIV programme managers, a 95% confidence level, and a 5% margin of error (e = 0.05), the calculation yielded a minimum sample size of 367 respondents. To account for anticipated non-response, the sample was increased by 10%, resulting in 408 questionnaires being prepared and distributed. The study also included 20 purposively selected key informants for the qualitative component, comprising senior programme managers, county HIV coordinators, and implementing partner representatives with relevant experience in donor-funded HIV programme management and implementation.

Sampling procedure: stratified random sampling was used to select the quantitative respondents. The target population was divided into three strata: implementing partner mechanisms, county HIV coordination structures overseeing donor-supported HIV services, and donor-supported health facilities. Based on the distribution of eligible senior programme managers in the sampling frame, the target sample of 408 respondents was allocated proportionately across the three strata (Table 1). The allocation comprised 9 respondents from implementing partner mechanisms, 6 from county HIV coordination structures, and 394 from donor-supported health facilities. Within each stratum, eligible senior programme managers were selected using simple random sampling. The sampling frame was developed from the PEPFAR PANORAMA Database (2024), which provided the institutional and managerial listings used to identify eligible respondents.

Variables: all items used a five-point Likert scale coded 1 = strongly agree, 2 = agree, 3 = neutral, 4 = disagree, and 5 = strongly disagree; hence, lower mean scores indicate stronger agreement and more favorable ratings. The explanatory variable was HRM strategies, measured using 25 statements across five dimensions: employee recruitment, employee compensation, training and development, employee integration, and employee utilization. An overall HRM score and dimension scores were calculated as unweighted means of the respective items and retained the 1-5 metric. The outcome variable was programme performance, measured using 20 statements across four domains: employee satisfaction, efficiency, country ownership and patient outcomes. An overall programme performance score and domain scores were similarly calculated as unweighted means. Programme performance was self-reported and perception based; the patient outcomes domain reflected managers' assessments of programme results. No routine health information system or programme records were abstracted to independently verify these outcomes.

Instrument development, validity and reliability: questionnaire items were developed from the HRM and related literature [9,13,16,25,26] and aligned with the study objectives and study constructs. Face and content validity were assessed through review by the supervising academic team, while a pilot study involving 40 respondents who were excluded from the main study was conducted to assess the clarity, comprehensibility, and suitability of the instrument. Internal consistency reliability was assessed using Cronbach's alpha during the pilot study. The 25-item HRM strategies scale achieved a Cronbach's alpha coefficient of 0.788, exceeding the 0.70 threshold for acceptable internal consistency and indicating that the items provided a sufficiently consistent measure of HRM strategies for subsequent composite scoring and analysis.

Data sources and measurement: primary quantitative data were collected using the structured, self-administered questionnaire described above, which captured respondents' demographic and professional characteristics, their ratings of HRM practices across the five dimensions, and their ratings of programme performance. The same administration procedures were used at all participating sites; participation was voluntary, and anonymity and confidentiality were assured. Qualitative data were collected concurrently through semi-structured key informant interviews with purposively selected programme managers, county HIV coordinators, and implementing partner representatives, using an interview guide covering human resource practices, the constraints managers encounter, and the perceived consequences of those constraints for HIV service delivery.

Mixed-methods integration: integration followed the convergent design [28,32], with quantitative and qualitative data collected during the same period in 2025 and analysed independently using appropriate procedures. Integration occurred during interpretation and reporting by comparing quantitative findings for each HRM dimension, programme performance domain, and the HRM-performance association with corresponding qualitative themes. Findings were classified as convergent when both sources pointed in the same direction, divergent when they differed, or expansive when qualitative findings provided information not captured quantitatively. These comparisons were presented in a joint display in the results, while integrated interpretations were considered in the discussion. Qualitative findings were used to contextualize and qualify the quantitative results rather than to independently confirm the correlation or regression estimates.

Bias: several potential sources of bias were considered. Selection bias was mitigated through stratified random sampling from a defined sampling frame, although the achieved sample was predominantly composed of managers working in Ministry of Health facilities, limiting assessment of representation across the sampling strata. Non-response bias could not be formally assessed because characteristics of non-respondents were not available. Information bias was addressed through use of a standardized, pilot-tested self-administered questionnaire and consistent administration procedures; however, all measures were based on respondents' perceptions. Common-method bias was possible because HRM strategies and programme performance were measured from the same respondents using the same instrument and response format, potentially contributing to the observed association [33]. Anonymity, confidentiality, voluntary participation, and standardized administration were used to reduce this risk, but no statistical test or adjustment for common-method variance was undertaken. Social desirability bias may also have resulted in more favorable assessments of organizational practices and performance, potentially inflating the observed scores and association. Recall bias was considered minimal because the questions focused primarily on current practices and programme conditions.

Statistical methods: quantitative data were analysed using the Statistical Package for the Social Sciences (SPSS) version 29. Descriptive statistics, including frequencies, percentages, means and standard deviations, were used to summarize respondent characteristics and item, dimension, domain and composite scores. Pearson correlation was used to quantify the strength and direction of the linear association between the composite HRM score and the composite programme performance score, with a 95% confidence interval obtained by Fisher's z transformation; because both composites are coded in the same direction, a positive coefficient indicates that more favorable HRM ratings accompany more favorable programme performance ratings. An unadjusted simple linear regression model, with programme performance as the dependent variable and the composite HRM score as the sole explanatory variable, was used to estimate the magnitude of that association; the coefficient is reported with its standard error, 95% confidence interval, t statistic and p value, and the model with R², adjusted R² and the F test. No covariates were included, and no multivariable model was fitted, so all estimates are unadjusted, and no confounding is controlled. Respondents were treated as independent observations; clustering within counties, organizations and facilities was not modelled, and neither cluster-robust standard errors nor multilevel models were fitted. Subgroup analyses by county, region or facility ownership were not undertaken, as the study was neither designed nor powered for subgroup inference. Statistical significance was assessed at p < 0.05, two-sided. Qualitative data were analysed thematically, with codes derived both from the interview guide and inductively from the transcripts and then grouped into themes [34], and were used to contextualize and qualify the quantitative findings rather than to test them [28,32].

Ethical considerations: ethical approval was obtained from the relevant university ethics review processes (USIU_A/ISERC/US987-2025), and authorization to conduct the study was granted by the National Commission for Science, Technology and Innovation (NACOSTI/P/25/4177314). Permission was also obtained from relevant county health authorities and implementing partner organizations. Participation was voluntary, informed consent was obtained from all respondents, and confidentiality was maintained throughout the study. No personally identifiable information was collected, and all data were stored securely and reported in aggregate form.

Results     

Participants: of the 408 questionnaires distributed to senior managers involved in the implementation and management of donor-funded HIV programmes across 40 counties in Kenya, 285 were completed, returned, and included in the analysis, giving a response rate of 70.5%; 119 questionnaires (29.5%) were not returned. No information was recorded for managers who did not respond, so respondents and non-respondents could not be compared. Male respondents constituted 169 of 285 (59.3%) and female respondents 116 (40.7%). Most respondents were aged between 25 and 44 years. Educational attainment was high: 142 respondents (49.8%) held a university or college degree, and a substantial further proportion held postgraduate qualifications. The largest group, 101 respondents (35.4%), had served in HIV programme management roles for between five and ten years, indicating considerable operational and managerial experience. By organizational affiliation, 203 respondents (71.2%) were based in Ministry of Health facilities, 53 (18.6%) in faith-based organizations and 29 (10.2%) in private-for-profit facilities. This affiliation variable describes the ownership of the facility or organization in which respondents worked, and not the three sampling strata; because the distribution of the target population across those strata was not documented, proportionate representation of implementing partner organizations, county HIV coordination structures and donor-supported health facilities cannot be demonstrated, and the achieved sample is most accurately described as one in which managers based in public-sector facilities predominate.

Implementation of HRM strategies: across the 25 items, the composite HRM score was 2.99 on the five-point scale (SD 0.86), indicating moderate institutionalization overall, with implementation varying between dimensions and between individual practices (Table 2). Items were coded 1 = strongly agree to 5 = strongly disagree, so lower means indicate stronger agreement. Employee recruitment attracted the strongest agreement (dimension mean 2.75), followed by employee utilization (2.95), employee integration (2.96), and training and development (3.00); employee compensation attracted the weakest agreement (3.27) and was the only dimension whose mean fell clearly on the disagreement side of the scale midpoint. Within recruitment, the conduct of competency-based interviews by panels with relevant technical expertise was the most strongly endorsed practice recorded anywhere in the instrument (M = 2.179, SD = 1.153), followed by pre-employment verification of licensure and references (M = 2.632, SD = 1.259) and written, accessible and consistently applied recruitment procedures (M = 2.793, SD = 1.223), while workload-based vacancy planning (M = 2.996, SD = 1.188) and digital applicant tracking (M = 3.144, SD = 1.266) were endorsed less strongly. Employee compensation was rated unfavorably throughout: benchmarking of salaries and benefits to national health sector and donor programme structures (M = 3.747, SD = 1.184) and the contribution of the compensation package to retention in hard-to-staff sites (M = 3.709, SD = 1.111) attracted the least agreement of any items in the instrument, and clarity of communication about salary bands, allowances and benefits (M = 3.158, SD = 1.300) and internal equity of remuneration across cadres, counties and facilities (M = 3.011, SD = 1.290) were also rated on the disagreement side; only the use of performance results to inform increments and non-financial recognition (M = 2.716, SD = 1.219) fell on the agreement side. Training and development, employee integration, and employee utilization each combined strongly and weakly endorsed practices. Appraisal-based individual development plans (M = 2.537, SD = 1.254) and annual training plans derived from documented competency gaps (M = 2.702, SD = 1.310) were endorsed more strongly than post-training mentorship and coaching (M = 3.112, SD = 1.257), observed training-related improvements in service quality and programme outputs (M = 3.042, SD = 1.280) and access to blended and accredited continuing professional development (M = 3.589, SD = 1.274), which was among the weakest-rated items overall. Within integration, shared targets and programme dashboards (M = 2.474, SD = 1.209), regular cross-functional meetings (M = 2.582, SD = 1.238) and structured induction of new staff (M = 2.649, SD = 1.268) were endorsed most strongly, whereas routine monitoring of integration quality (M = 3.407, SD = 1.288) and peer support and mentorship initiatives (M = 3.691, SD = 1.179) were endorsed least. Within utilization, prompt redeployment procedures for surges, absence and underutilization (M = 2.498, SD = 1.321) and the translation of utilization into improved coverage and fewer bottlenecks (M = 2.579, SD = 1.280) were endorsed more strongly than workload-based rostering (M = 3.112, SD = 1.234), skills-based task sharing (M = 3.109, SD = 1.253) and the use of human resource information systems and DHIS2 outputs to optimize deployment (M = 3.470, SD = 1.246). The pattern across the instrument is therefore one in which the procedural elements of workforce management, interview panels and pre-employment verification, structured induction, shared targets, appraisal-linked development planning and redeployment rules, attracted more agreement than the resource-dependent elements: pay and its benchmarking, retention packages, accredited continuing professional development, peer support and mentorship programmes, and the information systems used to guide deployment.

Programme performance: self-rated programme performance was measured by 20 items across four domains, with the same coding, so that lower means indicate stronger agreement (Table 3). For each item, Table 3 presents the percentage of the 285 respondents selecting each response option, with the corresponding frequency in parentheses. The composite programme performance score had a mean of 2.98 and a standard deviation of 0.90 on the five-point scale (n = 285). Ratings were uniformly mid-scale, and the four domains were closely spaced: employee satisfaction attracted the strongest agreement (domain mean 2.92), followed by country ownership (2.96), efficiency (3.01), and patient outcomes (3.01). No domain mean fell into either the strong-agreement or the weak-agreement band, and item means occupied a narrow range from 2.877 to 3.070. The distributions underlying these means were nevertheless widely dispersed rather than concentrated at the midpoint. Across the 20 items, an average of 37.9% of responses fell in the agree or strongly agree categories and 36.9% in the disagree or strongly disagree categories, with the remainder neutral, and item standard deviations ranged from 1.228 to 1.395. Managers were therefore divided about the performance of the programmes they manage, and the mid-scale composite reflects the balance between two substantial and opposed groups of respondents rather than a shared perception of moderate performance. Within employee satisfaction, clarity and manageability of job roles and workload attracted the strongest agreement of any item in the scale (M = 2.877, SD = 1.263), followed by supportive supervision incorporating coaching, fair appraisal and timely problem solving (M = 2.912, SD = 1.257) and support for work-life balance (M = 2.923, SD = 1.340); access to mentorship, continuing professional development and fair promotion attracted the least agreement in this domain (M = 2.968, SD = 1.290). Within efficiency, multidisciplinary collaboration reducing duplication and turnaround times (M = 2.975, SD = 1.344) and the contribution of electronic medical records and data tools (M = 2.989, SD = 1.315) were rated most favorably, whereas standard operating procedures and workflows enabling task completion within expected timelines attracted the least agreement (M = 3.039, SD = 1.395) and showed the widest dispersion of any item in the scale. Within country ownership, the items on leadership and priority setting, partner alignment through technical working groups, national information and reporting standards and use of domestic procurement and supply systems were rated almost identically (M = 2.954 to 2.968), and coordination of workforce decisions such as deployment, supervision and performance management through government systems attracted the least agreement (M = 2.975, SD = 1.243). Within patient outcomes, the domain rated least favorably overall, monitoring and active tracing of missed appointments and loss to follow-up (M = 3.070, SD = 1.246) and viral load testing coverage and suppression (M = 3.046, SD = 1.306) attracted the least agreement, while client experience (M = 2.954, SD = 1.228) and prompt linkage to care and treatment initiation (M = 2.975, SD = 1.333) were rated more favorably.

Qualitative accounts provided context for the quantitative ratings. Respondents identified training, mentorship, teamwork, and task shifting as important contributors to continuity of HIV services despite staffing constraints. One participant observed: "Training and mentorship have really improved how our staff handle viral load monitoring and patient follow-up. The teamwork is much better now than before." Another respondent stated: "Task shifting has really helped us manage patient flow. Even when we are understaffed, we reorganize ourselves and services continue."

These accounts indicate that managers attribute continuity of services in part to adaptive human resource practices. They describe perceived mechanisms rather than measured effects on service outputs, and the survey did not capture the service outputs to which managers referred. They also stand in contrast to the survey ratings of the corresponding items: access to blended and accredited continuing professional development (M = 3.589) and peer support and mentorship initiatives (M = 3.691) were among the least strongly endorsed items in the instrument. Interviewees were a small, purposively selected group describing mentorship and teamwork they had themselves experienced, whereas the survey asked all respondents whether such provision is systematically available. However, participants also highlighted concerns regarding recruitment governance and compensation systems. One respondent noted: "Recruitment is opaque. Sometimes we do not even sit in the interviews, yet we are expected to manage the staff afterward." Another explained: "We never understand how people are selected or how salaries are decided. The donor fixes everything."

These accounts converge with the quantitative ratings for employee compensation, which attracted the weakest agreement of any dimension in the survey and whose two least-endorsed items concerned the benchmarking of salaries and benefits and the adequacy of the package for retention in hard-to-staff sites. They diverge, however, from the quantitative ratings for employee recruitment, which attracted the strongest agreement of any dimension and whose most strongly endorsed item was the conduct of competency-based interviews by panels with relevant technical expertise.

Association between HRM strategies and programme performance: inferential analysis was undertaken to describe the direction, strength and statistical significance of the association between the composite HRM score and the composite programme performance score. Pearson correlation and unadjusted simple linear regression were used. These analyses quantify covariation between the two scores under the specified statistical model; the design does not permit inference about temporal direction or causal effect (Table 4).

Pearson correlation showed a positive and statistically significant association between HRM strategy scores and programme performance scores (r = 0.387, 95% CI 0.28 to 0.48; p < 0.001). Because both composites are coded in the same direction, with lower scores denoting stronger agreement, a positive coefficient means that programmes whose managers agreed more strongly that recruitment, compensation, training and development, integration and utilization practices are in place were also rated by those same managers as performing better. A coefficient of this magnitude is conventionally described as weak to moderate, and it describes the degree to which the two scores vary together rather than a demonstrated effect of one on the other.

In the unadjusted simple linear regression model, HRM strategy scores accounted for approximately 15% of the observed variance in programme performance scores in this sample (R² = 0.1499; adjusted R² = 0.1470; R = 0.3873). This figure describes variance explained within a single-predictor model estimated in these data; no judgement is offered as to whether it constitutes adequate explanatory power, since no established benchmark exists for this outcome, measure and setting.

The remaining variance cannot be attributed to any particular factor on the basis of these data. Organizational, institutional, financial and environmental characteristics that were not measured - including donor and funding arrangements, governance structures, institutional capacity, programme size, staffing establishment, health information systems and county context - may plausibly account for part of it, but none were measured or included in the model and their contribution was not estimated. The overall model was statistically significant (F (1, 283) = 49.94, p < 0.001) (Table 5). This indicates that a model including HRM strategy scores predicts programme performance scores better than a model without them, and the null hypothesis of no linear association between the two scores was therefore rejected. Rejecting that null hypothesis provides evidence of an association under the specified statistical model in these data.

The regression coefficient for HRM strategies was positive and statistically significant (β = 0.4028, SE = 0.0570, 95% CI 0.2906 to 0.5150, t = 7.0667, p < 0.001) (Table 6). A one-point difference in the HRM score was therefore associated with an estimated 0.4028-point difference in the same direction in the programme performance score on the same five-point scale; because lower scores denote stronger agreement, a one-point more favorable HRM rating accompanied a 0.4028-point more favorable programme performance rating. The intercept was 1.7744 (SE = 0.1771, 95% CI 1.4258 to 2.1230, t = 10.0167, p < 0.001). Based on the estimated coefficients, the fitted regression line is expressed as: programme performance = 1.7744 + 0.4028 (HRM).

The regression findings describe a positive association between more fully institutionalized HRM practice and higher-rated programme performance. They do not establish that strengthening recruitment systems, compensation practices, training and development, employee integration or workforce utilization would produce a corresponding improvement in programme performance. Qualitative accounts contextualized this association: respondents linked stronger teamwork, continuous staff mentorship, adaptive workforce deployment and supportive supervision to better service continuity, patient management and coordination across HIV service delivery points, while also emphasizing that weaknesses in recruitment governance, fragmented compensation decision-making and staffing instability constrain what programmes can achieve.

Integration of quantitative and qualitative findings: the joint display (Table 7) compares the principal quantitative findings with corresponding qualitative themes and classifies the relationships as convergent, divergent or expansive. The findings converged on employee compensation, which received the weakest agreement in the survey and was also a prominent concern in the interviews. The qualitative accounts similarly reflected the mid-range and mixed ratings of programme performance, describing continued service delivery despite operational constraints rather than uniformly strong performance. Divergence was observed for employee recruitment, which received the strongest agreement in the survey, while interviewees described recruitment processes as opaque and reported limited involvement in hiring decisions. This difference may reflect a distinction between the presence of recruitment procedures measured by the survey and managers' involvement in recruitment decisions explored qualitatively. Training and mentorship also showed divergence: interviewees described these practices as supporting service continuity, whereas items assessing accredited continuing professional development and peer support and mentorship were among the least strongly endorsed in the survey. The qualitative findings further expanded the quantitative results by identifying donor dependency, devolved governance arrangements and staffing instability as contextual factors not captured by the questionnaire. These factors provide additional context for the variation in programme performance not explained by the quantitative model.

Discussion     

Principal findings: among 285 senior managers of donor-funded HIV programmes across 40 Kenyan counties, HRM strategies were moderately institutionalized (composite mean 2.99), with employee recruitment receiving the strongest and employee compensation the weakest ratings. Programme performance was similarly rated at a mid-range level (composite mean 2.98, SD 0.90) across employee satisfaction, efficiency, country ownership and patient outcomes. Inferential analysis showed a significant positive association between HRM strategies and programme performance (r = 0.387, 95% CI 0.28 to 0.48; unadjusted β = 0.4028, 95% CI 0.29 to 0.52), with HRM strategies accounting for approximately 15% of the observed variation in programme performance. Thus, programmes with more favorable HRM ratings tended to have more favorable performance ratings. However, because the study was cross-sectional and both measures were reported by the same respondents, the findings indicate an association between perception-based measures and do not establish temporal direction or a causal effect of HRM strategies on programme performance.

Interpretation and comparison with previous evidence: the strongest ratings were observed for procedural aspects of HRM, including competency-based interview panels, pre-employment verification, structured induction, shared targets and dashboards, appraisal-linked development planning and redeployment arrangements. In contrast, lower ratings were observed for practices requiring recurrent financial resources or decisions beyond the programme level, including salary levels and benchmarking, retention packages for hard-to-staff sites, access to accredited continuing professional development, peer support and mentorship, and human resource information systems used to guide deployment. This pattern suggests that donor-funded HIV programmes may have established formal HRM procedures more readily than the resources and decision-making arrangements required to implement them consistently. Previous evidence indicates that supportive supervision, continuing professional development and structured mentoring contribute to health worker competence, motivation and service quality, while these functions may become constrained when programme financing is limited [16,21,25]. Evidence from HIV programmes in sub-Saharan Africa has similarly shown that task shifting can help maintain service coverage in settings affected by persistent shortages of health workers [7]. The divergence between the survey and qualitative findings regarding training and mentorship further suggests that such practices may be valued where available but are not consistently institutionalized across programmes.

Employee compensation received the weakest rating among the HRM dimensions, while qualitative findings indicated that the challenge extended beyond remuneration levels to the governance of employment and pay decisions. Interviewees described limited involvement in recruitment decisions, uncertainty regarding employment and salary determination, and concerns about differences in remuneration between donor-supported and government-employed staff working within the same service environment. Thus, while both sources identified compensation as an important concern, the qualitative findings provided additional insight into the distribution of decision-making authority over remuneration. This interpretation is consistent with evidence linking fragmented HRM governance, inequitable remuneration and limited career progression with dissatisfaction, reduced organizational commitment and staff turnover [20,35]. It also accords with evidence from Kenya showing that health sector planning and financial management responsibilities are distributed across national government, county governments and development partners [36]. The contrast between favorable ratings of recruitment procedures and qualitative concerns about control over recruitment further indicates the importance of distinguishing between the existence of an HRM practice and the authority to make decisions concerning its implementation.

Theoretical interpretation: the resource-based view proposes that organizational resources can generate sustained advantage when they are valuable, rare, difficult to imitate, and supported by organizational systems that enable their effective use [23,24]. The present findings suggest a distinction between formal HRM procedures and the human and organizational resources required to make them effective. The HRM practices that received stronger ratings, including structured recruitment, competency-based interviews, induction, appraisal, development planning, shared targets and redeployment, represent important organizational systems but are relatively accessible and can be replicated across programmes. In contrast, resources that develop through experience and social interaction, such as effective mentorship, continuing professional development, multidisciplinary collaboration, institutional knowledge and a stable and adequately supported workforce, are more difficult to develop and replicate. These areas received less favorable ratings, although interviewees who had access to mentorship described it as important for maintaining service continuity. From a resource-based view (RBV) perspective, therefore, the findings suggest that formal HRM structures alone may be insufficient to create strategically valuable human resources; sustained investment in employee development, supportive relationships, retention and organizational capacity is also required. The findings further suggest that organizational support for HRM may be constrained where key decisions concerning remuneration, staffing and resource allocation depend on actors outside individual programmes. Overall, the findings support the view that human resources and the systems through which they are developed and utilized are strategically important to programme performance [25,26].

Alternative explanations: the observed association between HRM strategies and programme performance may have several plausible explanations beyond a direct relationship between the two constructs. Because both measures were obtained from the same managers using the same instrument and response format, common method variance or respondents' general tendency to rate their organizations positively or negatively may have contributed to the association [33]. The direction of the relationship also cannot be established from the cross-sectional data. Programmes that are already performing well or have greater resources may have stronger HRM systems and may therefore receive more favorable HRM ratings, rather than stronger HRM practices necessarily preceding better performance. The association may also reflect differences in factors not included in the analysis, such as programme funding, maturity, facility ownership, staffing levels and county context. In addition, respondents were drawn from counties, organizations and facilities with potentially different operating environments, but this clustering was not accounted for in the analysis. Consequently, the reported standard errors and significance levels may not fully reflect the dependence among observations, although such clustering would not necessarily change the estimated direction or magnitude of the association.

Implications for donor-funded HIV programmes and for donor transition: the findings have important implications for the sustainability of donor-funded HIV programmes in Kenya, particularly as the HIV response faces changing levels and patterns of external financing [5,6,27]. The weaker ratings for remuneration and benchmarking, retention arrangements, accredited continuing professional development, mentorship and use of human resource information systems suggest areas that may require greater attention as responsibility for HIV services increasingly shifts towards national and county systems. The qualitative findings further indicate that fragmented responsibility for recruitment, remuneration and staffing decisions can limit programme managers' ability to address these challenges. Transition planning should therefore strengthen the integration of donor-supported HRM practices within existing national and county systems. Potential priorities include harmonizing donor-supported and government-funded positions within appropriate county remuneration structures, strengthening the involvement of programme managers in recruitment decisions, incorporating continuing professional development and structured mentorship into routine workforce plans and budgets, and improving the use of human resource and health information systems to inform staffing, deployment and service planning [36]. These measures should be evaluated using objective programme and workforce indicators to determine whether they improve service continuity and programme performance during donor transition.

Broader health system implications: the HRM functions examined in this study are relevant beyond HIV programming because they form part of routine health service management. The observed pattern of relatively stronger procedural HRM practices alongside weaker ratings for remuneration, professional development, mentorship and human resource information systems may therefore have relevance to other donor-supported health programmes and county health management. However, because the study was restricted to donor-funded HIV programmes, these broader implications remain hypotheses that require investigation in other health programmes and settings.

Strengths: the study included senior managers from donor-funded HIV programmes operating across 40 Kenyan counties and used a structured instrument covering five HRM dimensions and four programme performance domains. The quantitative component was complemented by purposively selected key informant interviews, enabling comparison between survey findings and managers' accounts of HRM practices and programme implementation. The study also examined both descriptive patterns and the statistical association between HRM strategies and programme performance, providing complementary quantitative and qualitative evidence.

Limitations: the cross-sectional design and reliance on self-reported measures from the same respondents limit causal interpretation and introduce the possibility of common method and social desirability bias. Programme performance was perception based, including the patient outcomes domain, and was not independently verified using routine programme or clinical indicators. The quantitative analysis did not account for clustering of respondents within counties, organizations or facilities, and the regression model was unadjusted for potential confounding factors such as programme characteristics, funding, staffing, and facility context. The study also excluded seven counties and was not designed for subgroup inference across counties or facility ownership categories. Finally, the qualitative component provides contextual explanations of the quantitative findings, but the available documentation does not permit full assessment against all COREQ reporting elements [30].

Implications for research: future research should examine HRM strategies alongside objective programme performance indicators and obtain explanatory and outcome measures from independent sources where possible. Longitudinal and quasi-experimental studies would help establish whether changes in HRM practices precede changes in programme performance, particularly during donor transition. Future studies should also account for clustering and relevant programme and organizational characteristics, report comprehensive reliability and construct validity evidence, and use adequately powered samples to examine differences across counties, facility types and organizational settings.

Conclusion     

In this cross-sectional mixed methods study, more favorable ratings of HRM strategies were significantly associated with more favorable self-reported programme performance among donor-funded HIV programmes in Kenya. The association was positive and of moderate magnitude (r = 0.387), with HRM strategies accounting for approximately 15% of the observed variation in programme performance. HRM strategies were moderately institutionalized, with stronger ratings for employee recruitment and weaker ratings for employee compensation, while programme performance was rated at a mid-range level across employee satisfaction, efficiency, country ownership and patient outcomes. The qualitative findings provided further context by highlighting constraints related to remuneration, professional development, mentorship, staffing and the broader governance of HRM. Because the study was cross-sectional and both constructs were measured through respondents' perceptions, the findings demonstrate association rather than causation and should not be interpreted as evidence that HRM strategies directly improve programme performance. Viewed through the resource-based view, the findings suggest that donor-funded HIV programmes have established important formal HRM structures, but that the resources and organizational conditions required to strengthen and sustain these practices remain uneven. Greater attention to remuneration, professional development, mentorship, staffing stability and the use of human resource information to support deployment may therefore be important as HIV programmes increasingly operate within national and county health systems. However, whether strengthening these areas leads to improved programme performance cannot be established from the present study and should be examined in future research using longitudinal designs and independent, objective programme performance indicators.

Competing interests     

The authors declare no competing interests.

Authors' contributions     

Mercy Njura Njeru conceptualized the study, developed the research design, conducted data collection, performed data analysis, interpreted the findings, and prepared the manuscript. All the authors contributed to the review of the study design, interpretation of results, and critical revision of the manuscript. All the authors read and approved the final version of this manuscript.

Acknowledgments     

The authors acknowledge all respondents who participated in this study, including programme managers, county HIV coordinators, implementing partner representatives, and health facility personnel. Appreciation is also extended to the institutions and stakeholders that facilitated data collection and provided support during the research process.

Tables     

Table 1: proportionate distribution of the target population and planned sample of senior HIV programme managers across donor-supported HIV programme structures in Kenya, 2025 (N = 4,488; n = 408)

Table 2: respondents' perceptions of human resource management strategies among senior managers of donor-funded HIV programmes operating in 40 donor-supported counties in Kenya, 2025 (n = 285)

Table 3: respondents' ratings of programme performance among senior managers of donor-funded HIV programmes operating in 40 donor-supported counties in Kenya, 2025 (n = 285)

Table 4: Pearson correlation coefficient between human resource management strategy scores and programme performance scores among donor-funded HIV programmes in Kenya, 2025 (n = 285)

Table 5: analysis of variance for the unadjusted simple linear regression model of the association between human resource management strategy scores and programme performance scores among donor-funded HIV programmes in Kenya, 2025 (n = 285)

Table 6: coefficients from the unadjusted simple linear regression of programme performance scores on human resource management strategy scores among donor-funded HIV programmes in Kenya, 2025 (n = 285)

Table 7: joint display integrating quantitative and qualitative findings on human resource management strategies and programme performance in donor-funded HIV programmes in Kenya, 2025

References