Digital health capacity and pediatric care quality in LMICs: A large-scale analysis of 5311 health facilities

Abstract

Objective

Digital health technologies are crucial for achieving universal health coverage (UHC), especially in low- and middle-income countries (LMICs) with limited digital infrastructure. This study aimed to assess digital health capacity across multiple LMICs and evaluate its association with evidence-based practice (EBP) and patient-centered care (PCC).

Methods

We analyzed Service Provision Assessment data collected over the past decade, spanning 5311 facilities and 20,880 pediatric visits across eight LMICs. Digital health capacity was measured using the WHO Classification of Digital Health Interventions (CDHI) across five domains: digital infrastructure, client engagement, healthcare providers, data services, and health system managers. EBP was assessed via ten binary items, while PCC was derived from eleven exit interview items using exploratory factor analysis. Multilevel regression models explored relationships between digital health capacity and both EBP and PCC.

Results

Overall digital health capacity was low (mean = 0.35), with notably low scores for digital infrastructure (0.02), healthcare providers (0.21), and health system managers (0.06). Digital health capacity was significantly associated with improved EBP (Coef. = 0.146, p < 0.001), particularly through digital infrastructure (Coef. = 0.183, p = 0.029), client engagement (Coef. = 0.205, p < 0.001), and provider capacity (Coef. = 0.142, p < 0.001). No significant effect emerged for PCC (Coef.=−0.013, p = 0.531).

Conclusions

The level of digital health technology in LMICs is generally insufficient, particularly in terms of digital infrastructure, healthcare provider training and health system managers. Although the implementation of digital health technologies has the potential to improve the EBP, its effect on enhancing PCC is relatively limited.

Keywords

Digital health evidence-based practice patient-centered care pediatric care low- and middle-income countries

Introduction

In 2022, an estimated 4.9 million children under five years of age died globally, with those in high-mortality countries facing an 80-fold higher risk compared to their counterparts in low-mortality settings.¹ Such disparities are closely linked to deficiencies in healthcare quality, limited access to healthcare services, and broader social determinants of health (SDOH)—including economic constraints, educational barriers, and infrastructural limitations—underscoring the need for a comprehensive and equity-focused approach to health system strengthening.^2–5 Enhancing digital health capabilities has emerged as a pivotal strategy in addressing this gap, offering a pathway to achieve the sustainable development goals (SDGs), including universal health coverage, as well as the World Health Organization's “triple billion” targets.^6,7

Digital health represents a transformative approach to healthcare delivery, leveraging information communication technology (ICT), data analytics, and computer science to broaden access, streamline services, and augment health information management.⁸ Its applications span telemedicine,⁹ wearable devices, electronic health records (EHRs), and advanced methodologies such as artificial intelligence (AI) and machine learning (ML).¹⁰ These innovations can bolster key functions in healthcare, including remote service provision, clinical decision support, and patient self-management.¹¹ The COVID-19 pandemic has further accelerated digital health adoption, particularly in low- and middle-income countries (LMICs), where constraints such as insufficient infrastructure or workforce shortages have prompted creative technological solutions.¹²

Despite growing interest, the influence of digital health technologies on overall healthcare quality remains inconsistently reported. While a number of studies have demonstrated beneficial outcomes,^12,13 others highlight marginal gains or even adverse effects related to patient safety and service quality.^14–16 Notably, the majority of this literature originates from high-income countries, where digital health has improved chronic disease management and care coordination.^17,18 However, evidence on pediatric healthcare quality in LMICs is comparatively sparse. Although some efforts have been made to reduce inequities,^19,20 persistent challenges include economic barriers, deficient infrastructure, and policy obstacles,²¹ compounded by limited training for healthcare professionals, weak stakeholder collaboration, and insufficient regulatory frameworks.^22,23 Such hurdles are evident in regions like sub-Saharan Africa and parts of the Middle East, exemplifying the urgent need for robust, context-specific digital health solutions.

This study aims to provide a comprehensive evaluation of digital health capacity in LMICs, particularly their impact on pediatric healthcare quality and user experience. We hypothesize that the implementation of digital health technologies will significantly enhance the efficiency and quality of pediatric healthcare services while improving the overall healthcare experience for patients and their families. The research will offer a scientific foundation for the future deployment of digital health solutions in LMICs, contributing to the development of evidence-based strategies that advance global child health, particularly in resource-constrained settings.

Methods

Data source

We used data from the service provision assessment (SPA), a nationally representative health facility survey implemented under the Demographic and Health Surveys (DHS) Program, funded by USAID.²⁴ The SPA provides comprehensive assessments of health service availability and quality in LMICs through standardized methodologies, enabling cross-country comparisons. Each SPA survey includes a nationally representative sample of health facilities from both public and private sectors and across all levels of care—from tertiary hospitals to community health posts. Multiple instruments are used for data collection, including facility inventories, direct clinical observations, provider interviews, and client exit interviews. These methods allow simultaneous assessment of structural readiness (e.g., infrastructure, equipment, diagnostics, and medicines) and process quality (e.g., adherence to clinical guidelines, patient-centered communication).²⁵

We selected countries that conducted SPA surveys involving sick child care observations and client exit interviews between 2014 and 2022. Where multiple datasets existed for a single country, only the most recent data were included. Our study comprises data from eight countries: Afghanistan (2019), the Democratic Republic of Congo (2018), Ethiopia (2022), Haiti (2018), Malawi (2014), Nepal (2021), Senegal (2019), and Tanzania (2015). In the pooled dataset, we derived key measures from multiple components of the SPA survey. The facility inventory module was used to construct the Digital Health Capacity Index, capturing the structural readiness of each facility. Clinical quality—operationalized as evidence-based practice—was assessed through direct observations of provider-client interactions during sick child consultations, where trained enumerators recorded provider adherence to standardized diagnostic and treatment protocols. Patient-centered care was measured using structured client exit interviews, which captured caregiver-reported experiences during the visit. Our study includes 5311 facilities and 20,880 sick children pooled for analysis: Afghanistan (84 facilities, 576 children), Democratic Republic of Congo (992, 2673), Ethiopia (813, 3755), Haiti (642, 2168), Malawi (748, 3441), Nepal (764, 2420), Senegal (253, 885), and Tanzania (1,015, 4962).

Measures and variables

Digital health capacity

We assessed digital health capacity at the facility level using the World Health Organization's Classification of Digital Health Interventions (CDHI), version 2.0.^8,26 The CDHI offers a globally recognized taxonomy for organizing and evaluating digital health functions, categorizing interventions by their primary users (e.g., clients, healthcare providers, system managers) and their intended functions (e.g., data services, communication, service delivery support). This classification provides a comprehensive and standardized framework to assess the structural readiness of health systems for digital transformation, particularly in low- and middle-income countries.²⁷

Guided by this framework and WHO technical guidance, we constructed a composite Digital Health Capacity Index (DHCI) using five domains: (1) digital infrastructure, (2) clients, (3) healthcare providers, (4) data services, and (5) health system managers (see Supplementary eTable 1). Each domain was operationalized through binary indicators from the Service Provision Assessment survey. For domains composed of multiple items (e.g., digital infrastructure and data services), conservative scoring rules were applied to ensure minimum functionality. The five domain scores were summed to construct a facility-level index ranging from 0 to 5. For visualization purposes (e.g., heatmaps), the score was rescaled to a 0–1 range by dividing by 5. The total DHCI score reflects the overall digital readiness of a facility, while individual domain scores capture specific functional capacities.

Following formative measurement principles,²⁸ we did not evaluate internal consistency, as each domain was conceptually distinct. As a robustness check, a principal component analysis (PCA) was conducted (Supplementary eFigure 1), and the PCA-derived score was highly correlated with the additive index (Pearson's r = 0.731, N = 20,342), supporting empirical validity.

Evidence-based practice

We assessed EBP using 10 binary clinical indicators derived from the WHO Integrated Management of Childhood Illness (IMCI) guidelines²⁹ and validated through the Good Medical Practice Index (GMPI) developed by Lewis et al.³⁰ using SPA data across LMICs. These indicators reflect core clinical activities (history-taking, physical examination, and caretaker counseling) required for appropriate diagnosis and management of sick children (see Supplementary eTable 2). All 10 items are routinely collected in SPA surveys and are aligned with international pediatric care standards. Validation studies have demonstrated high internal consistency, reliability, and construct validity, supporting their use as a robust indicator of clinical performance in LMICs.^30,31

To assess measurement robustness, we conducted a sensitivity analysis using an expanded 20-item quality tool developed by Macarayan et al.,³² which captures a broader range of clinical competencies. Results using this alternative definition of EBP were highly consistent with those from the 10-item index, supporting the stability and criterion validity of our primary measure.

Patient-centered care

PCC was measured using standardized patient-reported experience data from client exit interviews in the SPA survey. We initially selected 11 items spanning provider communication, responsiveness, respect for privacy, and facility conditions, each scored on a 3-point ordinal scale (1 = major problem, 2 = minor problem, 3 = no problem), with higher scores indicating better patient experience. Cronbach's alpha for the full 11-item scale was 0.79. To improve psychometric validity and reduce item redundancy, we conducted exploratory factor analysis (EFA) using principal axis factoring with oblique rotation. Seven items met the loading threshold (≥0.60) and formed a three-factor solution: communication, privacy/respect, and facility conditions (see Supplementary eTable 3). These dimensions aligned well with international frameworks including Consumer Assessment of Healthcare Providers and Systems (CAHPS^®),³³ Primary Care Assessment Tool (PCAT),³⁴ and WHO Framework on Integrated People-Centred Health Services (IPCHS),³⁵ supporting theoretical and construct validity (see Supplementary eTable 4).

The refined 7-item PCC scale (range: 3–21) demonstrated strong internal consistency (Cronbach's alpha = 0.78) and was used as the primary measure (see Supplementary eTable 5). Criterion-related validity was assessed using patient satisfaction as an independent variable in multilevel regression models, which showed a strong positive association (β = 0.929, 95% CI: 0.864 to 0.993, p < 0.001; see Supplementary eFigure 2).

Covariates

We assessed factors influencing clinical quality at the individual, provider, and facility levels, with covariates selected based on established determinants of healthcare quality.^24,30,36 Individual-level variables included the child's gender (male or female), age, and the caretaker's relationship (mother or non-mother). Provider-level variables comprised provider type (physician, nurse, or other), years of experience, weekly working hours, and the level of support, defined as having at least two of the following: a written job description, promotion opportunities, or non-monetary incentives. Facility-level variables included managing authority (government or private), location (urban or rural), and facility type (hospital or clinic).

Statistical analyses

The study commenced with a descriptive analysis of the characteristics of clients, healthcare providers, and health facilities. For continuous variables, means and standard errors were calculated, while categorical variables were summarized using frequencies and percentages. To ensure national representativeness across the eight countries, all descriptive analyses were weighted based on facility, healthcare worker, and client samples. Additionally, a heatmap was generated to visualize the levels of digital health capacity and its sub-dimensions across countries. To facilitate visual comparison across countries and sub- dimensions, descriptive summaries and heatmaps rescaled the index to a standardized range from 0 to 1. This rescaling allows for intuitive cross-national interpretation of digital readiness levels while maintaining analytical consistency.

A multilevel linear regression model was employed to examine the relationship between digital health capacity and the implementation of Evidence-Based Practices (EBP) and Patient-Centered Care (PCC). We used the unweighted additive Digital Health Capacity Index (0–5) in regression models, assigning one point per domain. This preserved effect size interpretability and followed standard formative index methodology. Initially, the total digital health capacity score was used as the independent variable to assess its association with clinical quality, both with and without control variables. Subsequently, digital health was divided into five sub-dimensions, each treated as an independent variable, to further investigate their relationship with clinical quality. Sensitivity analyses were conducted to ensure the robustness of the results and to confirm that the findings were consistent under different model specifications. To control for extreme values, Winsorization was applied to both the PCC and EBP metrics, with extreme values replaced by those corresponding to the 1st and 99th percentiles. All regression analyses were performed without weighting. Statistical analyses were performed using Stata version 17.

Results

Table 1 shows the demographics and characteristics of facilities, health providers, caretakers, and children in the digital health study across eight countries. The study analyzed 20,699 observations from Afghanistan (2.77%, n = 574), the Democratic Republic of Congo (12.85%, n = 2660), Ethiopia (18.08%, n = 3742), Haiti (10.46%, n = 2166), Malawi (16.08%, n = 3329), Nepal (11.51%, n = 2383), Senegal (4.27%, n = 884), and Tanzania (23.97%, n = 4961). Out of 4705 surveyed facilities, 35.99% (n = 1693) were urban and 64.01% (n = 3012) were rural. Management-wise, 66.86% (n = 3146) were government-run, while 33.14% (n = 1559) were privately managed. Clinics were constituted 89.54% (n = 4213) of the facilities, with hospitals comprising 10.46% (n = 492). The dataset included 7360 healthcare providers: physicians (39.17%, n = 2883), nurses (38.75%, n = 2852), and others (22.08%, n = 1625). Providers averaged 6.24 years of experience (SD = 0.14), worked 50.84 h per week (SD = 0.58), and had 15.47 years of education (SD = 0.10). The support environment scored a mean of 0.34 (SD = 0.01). Among 20,694 caretakers, 81.94% (n = 16,957) were mothers and 18.06% (n = 3737) were non-mothers. The children's mean age was 20.38 months (SD = 0.18), with a gender distribution of 52.91% male (n = 10,951) and 47.09% female (n = 9748).

Table 1.

Characteristics of the facility, provider, caretaker, and sick child.

Characteristics	Weighted sample	Value
Country, No. (%)	20,699
Afghanistan		574	2.77
Congo Democratic Republic		2660	12.85
Ethiopia		3742	18.08
Haiti		2166	10.46
Malawi		3329	16.08
Nepal		2383	11.51
Senegal		884	4.27
Tanzania		4961	23.97
Facility
Location, No. (%)	4705
Urban		1693	35.99
Rural		3012	64.01
Manage authority, No. (%)	4705
Government		3146	66.86
Private		1559	33.14
Type, No. (%)	4705
Hospital		492	10.46
Clinic		4213	89.54
Provider
Type, No. (%)	7360
Physician		2883	39.17
Nurse		2852	38.75
Others		1625	22.08
Years of working, mean (SD)	7728	6.24	0.14
Working hours per week, mean (SD)	7728	50.84	0.58
Years of education, mean (SD)	7728	15.47	0.1
Support environment, mean (SD)	7728	0.34	0.01
Caretaker
Child relationship, No. (%)	20,694
Mother		16,957	81.94
Non-mother		3737	18.06
Child
Age of the month	19,950	20.38	0.18
Gender	20,699
Male		10,951	52.91
Female		9748	47.09

Note: Values are presented as n (%) for categorical variables and mean (standard deviation) for continuous variables. All statistics are weighted using the Service Provision Assessment (SPA) survey weights, which account for the complex sampling design across facilities, healthcare providers, and client levels to ensure national representativeness.

The heatmap is presented in Figure 1. The horizontal axis denotes the number of countries used in our study whereas, the vertical axis represents the total digital health score and the scores for each category. Red color indicates lower scores, while blue indicates higher scores. From the figure, it can be seen that the average digital health score across LMICs is 0.35. The scores for the five sub-dimensions are as follows: 0.02 for digital infrastructure, 0.52 for client engagement, 0.21 for healthcare providers, 0.92 for data services, and 0.06 for health system managers. The country with the highest total digital health score is Senegal, with a score of 0.39. Scores for digital infrastructure and health system managers are consistently low across all the countries however; the scores for clients and healthcare providers are moderate. The overall scores for data service are consistently high among each country, precisely: Afghanistan (0.59), Democratic Republic of the Congo (0.99), Ethiopia (0.69), Haiti (0.92), Malawi (0.89), Nepal (0.97), Senegal (0.95), and Tanzania (0.92).

Figure 1.

National variation in digital health capacity and sub-dimensions across eight LMICs. Note: Heatmap showing the mean facility-level scores for digital health capacity and its five functional domains across eight LMICs. Data are weighted and derived from SPA surveys (2014–2022). Digital Health scores are rescaled from 0–5 to 0–1 for visualization purposes. Original regression analyses were conducted using the unstandardized total score (range: 0–5). “Mean” represents the average across countries.

Table 2 shows the multilevel linear regression results assessing the relationship between digital health capacity and EBP. We have found positive and significant associations for total digital health (β=0.146, P < 0.001); digital infrastructure, 0.183 (P = 0.029); clients 0.205 (P < 0.001); and health providers 0.142 (P < 0.001). For covariates, the support environment also found a strong positive association with coefficients of 0.204 (P < 0.001). Conversely, the negative associations were identified for the age of the sick child −0.006 (P < 0.001), and non-mother caretakers −0.107 (P < 0.001). Other provider types also highlighted the negative association of −0.288 (P < 0.001), and years of working found a negative association of −0.007 (P = 0.005). Facility characteristics further influenced results, with rural facilities exhibiting positive associations of 0.168 (P < 0.001), whereas clinics, compared to hospitals, showed a negative impact of −0.104 (P = 0.049). Our results highlighted the complex impact of digital health components on clinic quality, highlighting the critical roles of digital infrastructure and supportive environments in enhancing healthcare outcomes.

Table 2.

Associations between digital health capacity and evidence-based practice: multilevel linear regression results.

Variable	M1 Coef.	95%CI	M1a Coef.	95%CI	M2 Coef.	95%CI	M2a Coef.	95%CI
Digital health (total) ^a	0.146***	[0.10,0.19]	0.111***	[0.07,0.16]
Digital infrastructure^a					0.183*	[0.02,0.34]	0.13	[−0.04,0.30]
Clients^c					0.205***	[0.12,0.29]	0.173***	[0.09,0.26]
Health providers^b					0.142***	[0.06,0.23]	0.122**	[0.04,0.21]
Data services^a					0.026	[−0.13,0.18]	−0.057	[−0.21,0.10]
Health system managers^a					0.111	[−0.02,0.24]	0.083	[−0.05,0.22]
Child
Gender (Ref. = Female)^c			0.004	[−0.04,0.05]			0.007	[−0.03,0.05]
Age of sick child (month)^c			−0.006***	[−0.01−−0.00]			−0.006***	[−0.01,−0.00]
Caretaker
Non-mother (Ref. = Mother)^c			−0.107***	[−0.17,−0.05]			−0.104***	[−0.17,−0.04]
Age of caretaker (year)^c			−0.002	[−0.01,0.00]			−0.002	[−0.01,0.00]
Provider
Nurse (Ref. = Physician)^b			−0.03	[−0.13,0.07]			−0.014	[−0.11,0.09]
Others			−0.288***	[−0.40,−0.18]			−0.278***	[−0.39,−0.17]
Years of working^b			−0.007**	[−0.01,−0.00]			−0.008**	[−0.01,−0.00]
Working hours per day^b			0	[−0.00,0.00]			0	[−0.00,0.00]
Years of education^b			0.004	[−0.01,0.02]			0.005	[−0.01,0.02]
Support environment^b			0.204***	[0.14,0.27]			0.205***	[0.14,0.27]
Facility
Private (Ref. = Government)^a			0.102*	[0.01,0.19]			0.084	[−0.01,0.17]
Rural (Ref. = Urban)^a			0.168***	[0.08,0.25]			0.191***	[0.10,0.28]
Clinic (Ref. = Hospital)^a			−0.104*	[−0.20,−0.00]			−0.117*	[−0.22,−0.01]

Note: Model (M) 1 and M1a examine the association between the total digital health score and EBP, with M1a adjusting for covariates. M2 and M2a assess individual digital health domains, with M2a including full covariate adjustment.

represents facility level; ^brepresents healthcare provider level; ^crepresents client level.

* P < 0.05, ** P < 0.01, *** P < 0.001

Coefficient values such as 0.000 or −0.000 indicate very small but non-zero effects;

Confidence intervals such as [−0.01, 0.00] reflect small estimated effects near zero, rounded to two decimal places. These do not imply exact zero bounds.

Table 3 presents the results of the multilevel linear regression analysis examining the relationship between digital health capacity and PCC. The analysis showed that the regression coefficient of total digital health on PCC was negative but non-significant at −0.013 (P = 0.531). After controlling for confounding factors, the coefficient value was positive but still, it was non-significant at 0.044 (P = 0.098). Digital infrastructure also showed a negative relationship before and after controlling for confounding factors at −0.157 (P = 0.076) and −0.091 (P = 0.321), respectively. Clients showed a significant negative association of −0.136 (P = 0.006), but after controlling for confounding factors, a non-significant association of −0.08 (P = 0.119) was observed. Health providers have demonstrated positive and significant relationships before and after controlling for confounding factors with 0.122 (P = 0.023) and 0.115 (P = 0.034). Data services found a non-significant positive association before controlling for confounding factors, with a coefficient of 0.046 (P = 0.604), while controlling for confounding factors found a significant positive association of 0.210 (P = 0.023). Other provider types showed positive associations with 0.179 (P = 0.005), and years of education also had positive relationship of 0.021 (P = 0.005). The private facilities exhibiting positive associations of 0.552 (P < 0.001).

Table 3.

Associations between digital health capacity and patient-centered care: multilevel linear regression results.

Variable	M3 Coef.	95%CI	M3a Coef.	95%CI	M4 Coef.	95%CI	M4a Coef.	95%CI
Digital health (total) ^a	−0.013	[−0.06,0.04]	0.044	[−0.01,0.10]
Digital infrastructure ^a					−0.157	[−0.34,0.02]	−0.091	[−0.28,0.09]
Clients ^c					−0.136**	[−0.23,−0.04]	−0.08	[−0.18,0.02]
Health providers ^b					0.122*	[0.02,0.22]	0.115*	[0.01,0.22]
Data services ^a					0.046	[−0.13,0.22]	0.210*	[0.03,0.39]
Health system managers ^a					0.107	[−0.04,0.25]	0.117	[−0.04,0.27]
Child ^c
Gender (Ref. = Female) ^c			0.008	[−0.05,0.06]			0.007	[−0.05,0.06]
Age of sick child (month) ^c			−0.002*	[−0.00,−0.00]			−0.002	[−0.00,0.00]
Caretaker ^c
Non-mother (Ref. = Mother) ^c			−0.038	[−0.12,0.04]			−0.047	[−0.13,0.03]
Age of caretaker (year) ^c			0.001	[−0.00,0.00]			0	[−0.00,0.00]
Provider ^b
Nurse (Ref. = Physician) ^b			−0.083	[−0.20,0.03]			−0.103	[−0.22,0.02]
Others			0.179**	[0.05,0.31]			0.150*	[0.02,0.28]
Years of working ^b			0.005	[−0.00,0.01]			0.005	[−0.00,0.01]
Working hours per day ^b			0	[−0.00,0.00]			0	[−0.00,0.00]
Years of education ^b			0.021**	[0.01,0.04]			0.019*	[0.00,0.03]
Support environment ^b			0.044	[−0.04,0.13]			0.044	[−0.04,0.13]
Facility ^a
Private (Ref. = Government) ^a			0.552***	[0.45,0.65]			0.579***	[0.48,0.68]
Rural (Ref. = Urban) ^a			0.073	[−0.02,0.17]			0.037	[−0.07,0.14]
Clinic (Ref. = Hospital) ^a			0.347***	[0.23,0.46]			0.345***	[0.23,0.46]

Note: Model (M)3 and M3a examine the association between the total digital health score and PCC, with M3a adjusting for covariates. M4 and M4a assess individual digital health domains, with M4a including full covariate adjustment.

represents facility level; ^b represents healthcare provider level; ^c represents client level.

* P < 0.05, ** P < 0.01, *** P < 0.001

Coefficient values such as 0.000 or −0.000 indicate very small but non-zero effects;

Confidence intervals such as [−0.00, 0.00] reflect small estimated effects near zero, rounded to two decimal places. These do not imply exact zero bounds.

Table 4 illustrates the results of the sensitivity analysis between digital health capacity, EBP, and PCC (see eTables 6–9 in Supplement for details). For the EBP, total digital health demonstrated positive and significant connotations before and after controlling for confounding factors at 0.249 (P < 0.001) and 0.184 (P < 0.001) respectively. The winsorized model found the relationship at 0.145 (P < 0.001) and 0.110 (P < 0.001). Healthcare providers and clients also demonstrated a significant positive relationship before and after controlling for confounding factors.

Table 4.

Sensitivity analyses of the associations between digital health and EBP and PCC outcomes.

Variable	M5 Coef.	95%CI	M5a Coef.	95%CI	M6 Coef.	95%CI	M6a Coef.	95%CI
EBP-20 items
Digital health	0.249***	[0.18,0.32]	0.184***	[0.11,0.26]
Digital infrastructure					0.192	[−0.06,0.45]	0.091	[−0.17,0.35]
Clients					0.399***	[0.27,0.53]	0.325***	[0.19,0.46]
Healthcare providers					0.299***	[0.17,0.42]	0.241***	[0.11,0.37]
Data services					0.227	[−0.01,0.47]	0.056	[−0.19,0.30]
Health system managers					0.003	[−0.20,0.21]	−0.017	[−0.23,0.20]
EBP 1^st-99^th winsorization
Digital health (total)	0.145***	[0.10,0.19]	0.110***	[0.06,0.16]
Digital infrastructure					0.182*	[0.02,0.34]	0.128	[−0.04,0.29]
Clients					0.203***	[0.12,0.29]	0.171***	[0.09,0.26]
Healthcare providers					0.140***	[0.06,0.22]	0.119**	[0.03,0.20]
Data services					0.028	[−0.12,0.18]	−0.055	[−0.21,0.10]
Health system managers					0.11	[−0.02,0.24]	0.081	[−0.05,0.22]
PCC-11 items
Digital health (total)	−0.108***	[−0.19, −0.03]	0.039	[−0.04, 0.12]
Digital infrastructure					−0.456**	[−0.74, −0.17]	−0.240	[−0.53, 0.05]
Clients					−0.306***	[−0.46, −0.15]	−0.162*	[−0.32, −0.01]
Healthcare providers					0.173*	[0.01, 0.33]	0.209*	[0.05, 0.37]
Data services					−0.150	[−0.43, 0.13]	0.152	[−0.13, 0.43]
Health system managers					0.122	[−0.11, 0.36]	0.162	[−0.08, 0.40]
PCC 1^st-99^th winsorization
Digital health (total)	−0.012	[−0.06,0.04]	0.043	[−0.01,0.09]
Digital infrastructure					−0.152	[−0.33,0.02]	−0.088	[−0.27,0.09]
Clients					−0.133**	[−0.23,−0.04]	−0.078	[−0.17,0.02]
Healthcare providers					0.124*	[0.03,0.22]	0.117*	[0.02,0.22]
Data services					0.029	[−0.14,0.20]	0.187*	[0.01,0.36]
Health system managers					0.103	[−0.04,0.25]	0.111	[−0.04,0.26]

Note: Model (M) 5 evaluates the association between the total Digital Health Capacity Index and each outcome: (i) EBP measured by 20 items; (ii) EBP with winsorized values; (iii) PCC measured using a validated 11-item score; (iv) PCC with winsorized values, and Model 5a includes covariates. M6 assesses each digital health domain, and M6a includes covariates.

* P < 0.05, ** P < 0.01, *** P < 0.001

For PCC, results remained consistent across different specifications. In the PCC-11 items model, total digital health showed a significant negative association before covariate adjustment (−0.108, P = 0.006), which became non-significant after adjustment (0.039, P = 0.300). Digital infrastructure and clients also demonstrated negative relationships in both models, with digital infrastructure showing a significant association (−0.456, P = 0.002) before adjustment. In contrast, healthcare providers were positively associated with PCC, with coefficients of 0.173 (P = 0.036) and 0.209 (P = 0.012) before and after controlling for covariates, respectively. Similarly, in the winsorized PCC model, total digital health had a non-significant connection in both models −0.012 (P = 0.598) and 0.043 (P = 0.101) correspondingly. Digital infrastructure and clients also highlighted the negative associations in both models, with one significant association for clients −0.133 (P = 0.006). Health providers have positive and significant links 0.124 (P = 0.020) and 0.117 (P = 0.022) before and after controlling for confounding factors.

Additionally, Supplement eTable 10 presents sensitivity analyses adjusting for pediatric consultation volume. The results confirm that the association between digital health and EBP remained significant and robust (Coef. = 0.116, P < 0.001) after accounting for facility-level variation in child caseload. Associations with PCC and individual digital health domains were also consistent in magnitude and direction, supporting the stability of our main findings across different facility types.

Discussion

Our study characterizes the foundational landscape of digital health technology in LMICs and examines its influence on influence on evidence-based practice and patient-centered care. Drawing on validated, multidimensional constructs of digital health capacity, EBP, and PCC, we found that digital health capacity—though underdeveloped in many LMICs—was significantly associated with improvements in clinical performance (EBP), but had limited influence on patient experience (PCC).

These results align with previous studies. For example, Mansah et al., reported positive outcomes in seven of nine dimensions related to consumer experiences post-digitalization.³⁷ Similarly, Ndayishimiye et al.'s comprehensive review during the COVID-19 pandemic revealed that digital health tools were pivotal in facilitating virtual healthcare, providing clinical assistance, monitoring care quality, and managing critical health resources.^38,39 Unlike most prior research focused on chronic care in high-income countries with robust ICT infrastructure (e.g., Australia, Canada, the UK, and the US),^17,18 our study centers on pediatric care in LMICs and uses harmonized measures to enable cross-country comparison.

Our domain-specific analysis offers additional insight. Digital infrastructure was positively associated with EBP, reinforcing the importance of stable internet connectivity, reliable electricity, and computing equipment for digital interventions.^39–41 These results echo prior findings from China and elsewhere emphasizing the foundational role of digital infrastructure in expanding access to care and enabling advanced tools like telemedicine and electronic health records.⁴²

In terms of client interactions, —such as patient feedback systems—also showed strong positive associations with EBP, suggesting that systems which incorporate patient voice can support higher quality clinical encounters. These results are consistent with literature underscoring the value of feedback loops in quality improvement and patient-centered systems.^43,44 Studies, such as those by Rachel et al., demonstrate that incorporating patient perspectives significantly enhances patient experiences and identifies areas for service improvement.⁴⁵ Hiyam et al. propose that to enhance service quality, control systems should prioritize the dissemination phase of patient survey results among healthcare professionals.⁴⁶ However, we also found a negative association between client feedback tools and PCC, indicating that implementation of digital tools may not automatically translate into a better patient experience. Possible explanations include superficial adoption, limited patient engagement, or increased administrative burden on providers.⁴¹

Among healthcare providers, training and digital support systems were robustly associated with improved EBP.^47,48 These findings affirm the importance of equipping providers with the tools and competencies to navigate digital systems and leverage health information for clinical decision-making. In line with existing literature, facilities with trained providers using health management information systems (HMIS) were better positioned to deliver guideline-based care.⁴⁹

The associations between data systems and EBP were weaker and inconsistent, highlighting potential challenges in the quality or usage of digital data systems in LMICs. While routine data reporting is essential for performance monitoring, its direct clinical utility may depend on real-time accessibility, provider capacity, and integration with clinical workflows.⁵⁰ Similarly, digital capacity among health system managers—such as inventory tracking and supervision—showed limited direct associations with care outcomes, possibly reflecting the structural rather than frontline nature of these functions. Studies have demonstrated that effective inventory management using digital health tools significantly improves operational efficiency and healthcare outcomes by ensuring timely access to medications and optimizing the use of financial and human resources.⁵¹ However, in low-income countries, the absence of robust data collection and reporting systems, coupled with limited adoption of digital health tools like inventory management systems, hampers their impact on healthcare quality.^52,53

Notably, total digital health capacity was not significantly associated with patient-centered care (PCC) across most model specifications. This finding is consistent with prior literature suggesting that digitalization alone may not translate into improved patient-reported experiences, particularly in resource-constrained settings characterized by low health literacy, limited patient engagement, and contextual factors such as long wait times or poor facility conditions.⁵⁴ One plausible explanation is that PCC is inherently shaped by patient expectations, which may be modest in low- and middle-income country contexts. Consequently, a positive user experience may be reported despite limited digital integration, and conversely, higher digital capacity does not guarantee improved perception if these contextual factors are not addressed. Moreover, environmental elements such as cleanliness, noise levels, and the availability of basic amenities (e.g., food provision) have been shown to significantly influence user experience in hospital settings.⁵⁵ Core components of optimized PCC include timely access to care, patient engagement, a strong patient-physician relationship, evidence-based care, clear communication, physical and emotional support, family involvement, personalized care, service responsiveness, and continuity of care.^56–58 While digital health tools may enhance some of these domains—such as communication or follow-up tracking—their effectiveness likely depends on how they are integrated with broader care delivery systems and whether they are accompanied by institutional efforts to enhance responsiveness and human-centered service design.

Despite the significant findings, our study has several limitations. First, the cross-sectional design restricts our ability to draw causal inferences regarding the observed associations. Second, although the digital health, EBP, and PCC measures were rigorously developed using validated indicators and frameworks, they were inherently limited by the scope of available SPA data and may not capture the full complexity of digital functionality or patient experience. Third, while the dataset covers eight LMICs, findings may not be generalizable to countries with substantially different health system structures or levels of digital health maturity. Finally, despite comprehensive sensitivity analyses, future research should incorporate longitudinal designs and mixed-method approaches to better understand underlying mechanisms and long-term effects.

Conclusion

Our study demonstrates that digital health capacity significantly improve EBP in LMICs, particularly in relation to digital infrastructure, patient (feedback systems), and provider (training). Its influence on patient experience is more nuanced and may require complementary strategies to translate digital investment into perceived care improvements. These findings underscore the need for comprehensive approaches that address digital infrastructure, training, and patient engagement to optimize the benefits of digital health in enhancing pediatric healthcare services in resource-limited settings.

Supplemental Material

sj-docx-1-dhj-10.1177_20552076251349890 - Supplemental material for Digital health capacity and pediatric care quality in LMICs: A large-scale analysis of 5311 health facilities

Supplemental material, sj-docx-1-dhj-10.1177_20552076251349890 for Digital health capacity and pediatric care quality in LMICs: A large-scale analysis of 5311 health facilities by Dan Wang, Zhongliang Zhou, Mengyao Li and Wenhua Wang in DIGITAL HEALTH

Footnotes

Acknowledgments

The authors would like to thank all the participants involved in this study.

ORCID iD

Dan Wang

Ethical considerations

The original survey implementers obtained ethical approvals for data collection. Procedures and questionnaires for standard DHS surveys were reviewed and approved by the ICF Institutional Review Board (IRB). This secondary analysis was exempted from human subjects review as it exclusively utilized anonymized data available in the public domain.

Author contributions

Dan Wang and Wenhua Wang conceived and designed the study. Mengyao Li conducted the data analysis and interpreted the results. Dan Wang drafted the manuscript with support from Wenhua Wang and Zhongliang Zhou. All authors reviewed and approved the final version for publication.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data sharing statement

The data underlying this study are available in Service Provision Assessment (SPA) surveys at .

Supplemental material

Supplemental material for this article is available online.

References

Estimation UNI-aGfCM. Levels & Trends in Child Mortality: Report 2022, estimates developed by the United Nations Inter-agency Group of Child Mortality Estimation. New York, NY: United Nations Children's Fund, 2024.

Braveman

Gottlieb

. The social determinants of health: it's time to consider the causes of the causes. Public Health Rep 2014; 129: 19–31.

Kruk

Lewis

Arsenault

, et al. Improving health and social systems for all children in LMICs: structural innovations to deliver high-quality services. Lancet 2022; 399: 1830–1844.

Marmot

Allen

Bell

, et al. WHO European review of social determinants of health and the health divide. Lancet 2012; 380: 1011–1029.

Solar

Irwin

. A conceptual framework for action on the social determinants of health. Geneva, Switzerland: WHO Document Production Services, 2010.

Kickbusch

Piselli

Agrawal

, et al. The Lancet and Financial Times Commission on governing health futures 2030: growing up in a digital world. Lancet 2021; 398: 1727–1776.

Michael McGinnis

Harvey

Fineberg

VJD

. Advancing the learning health system. N Engl J Med 2021; 385: –5.

WHO. Guideline recommendations on digital interventions for health system strengthening. 2019.

Collins

Akselrod

Altymysheva

, et al. The promise of digital health technologies for integrated care for maternal and child health and non-communicable diseases. Br Med J 2023; 381: e071074.

10.

Cuff

. The evolution of digital health and its continuing challenges. BMC Digital Health 2023; 1: 3.

11.

Tahsin

Armas

Kirakalaprathapan

, et al. Information and communications technologies enabling integrated primary care for patients with complex care needs: scoping review. J Med Internet Res 2023; 25: e44035–20230419.

12.

Hollander

Carr

. Virtually perfect? Telemedicine for COVID-19. N Engl J Med 2020; 382: 1679–1681.

13.

Menachemi

Chukmaitov

Saunders

, et al. Hospital quality of care: does information technology matter? The relationship between information technology adoption and quality of care. Health Care Manage Rev 2008; 33: 51–59.

14.

Roth

Morton

, et al. Systematic review: impact of health information technology on quality, efficiency, and costs of medical care. Ann Intern Med 2006; 144: 742–752.

15.

Del Beccaro

Jeffries

Eisenberg

, et al. Computerized provider order entry implementation: no association with increased mortality rates in an intensive care unit. Pediatrics 2006; 118: 290–295.

16.

Koppel

Metlay

Abaluck

, et al. Role of computerized physician order entry systems in facilitating medication errors. JAMA 2005; 293: 1197–1203.

17.

Riordan

McHugh

O’Donovan

, et al. The role of physician and practice characteristics in the quality of diabetes management in primary care: systematic review and meta-analysis. J Gen Intern Med 2020; 35: 1836–1848.

18.

Kallas

Sandhu

Gandilo

, et al. Use of digital health technology in heart failure and diabetes: a scoping review. J Cardiovasc Transl Res 2023; 16: 526–540.

19.

Colicchio

Cimino

. Twilighted homegrown systems: the experience of six traditional electronic health record developers in the post-meaningful use era. Appl Clin Inform 2020; 11: 356–365.

20.

Ngugi

Babic

Kariuki

, et al. Development of standard indicators to assess use of electronic health record systems implemented in low-and medium-income countries. PLoS One 2021; 16: e0244917.

21.

Irshad

Mehrun

Ghafoor

. Infrastructure and economic growth: evidence from lower middle-income countries. J Knowl Econ 2023; 14: 161–179.

22.

Awol

Birhanu

Mekonnen

, et al. Health professionals’ readiness and its associated factors to implement electronic medical record system in four selected primary hospitals in Ethiopia. Adv Med Educ Pract 2020; 11: 147–154.

23.

Essuman

Apaak

Ansah

, et al. Factors associated with the utilization of electronic medical records in the Eastern Region of Ghana. Health Policy Technol 2020; 9: 362–367. DOI:10.1016/j.hlpt.2020.08.002

24.

Moucheraud

McBride

. Variability in health care quality measurement among studies using service provision assessment data from low- and middle-income countries: a systematic review. Am J Trop Med Hyg 2020; 103: 986–992.

25.

Program

. SPA Overview, https://www.dhsprogram.com/methodology/Survey-Types/SPA.cfm (accessed June 2024).

26.

WHO. Classification of digital interventions, services and applications in health. 2023.

27.

Pernencar

Aguilar

Saboia

, et al. Systematic mapping of digital health apps–A methodological proposal based on the World Health Organization classification of interventions. Digital Health 2022; 8: 20552076221129071.

28.

Bollen

Diamantopoulos

. In defense of causal-formative indicators: a minority report. Psychol Methods 2017; 22: 581.

29.

Organization

. Integrated Management of childhood Illness (chart booklet). Geneva, Switzerland: World Health Organization, 2008.

30.

Lewis

McConnell

Aryal

, et al. Health service quality in 2929 facilities in six low-income and middle-income countries: a positive deviance analysis. Lancet Glob Health 2023; 11: e862–e870.

31.

Lewis

Roder-DeWan

Malata

, et al. Clinical performance among recent graduates in nine low-and middle-income countries. Trop Med Int Health 2019; 24: 620–635.

32.

Macarayan

Gage

Doubova

, et al. Assessment of quality of primary care with facility surveys: a descriptive analysis in ten low-income and middle-income countries. Lancet Glob Health 2018; 6: e1176–e1185.

33.

Orr

Zaslavsky

Hays

, et al. Development, methodology, and adaptation of the Medicare Consumer Assessment of Healthcare Providers and Systems (CAHPS®) patient experience survey, 2007–2019. Health Serv Outc Res Methodol 2023; 23: 1–20.

34.

Shi

Starfield

. Validating the adult primary care assessment tool. J Fam Pract 2001; 50: 161.

35.

Hafiz

Yin

Sun

, et al. Examining the use and application of the WHO integrated people-centred health services framework in research globally–a systematic scoping review. Int J Integr Care 2024; 24: 9.

36.

Lee

Cooper

Kruk

. Regional determinants of quality of care for sick children: a multilevel analysis in four countries. J Glob Health 2024; 14: 04053.

37.

Preko

Budu

. The impact of digital health technologies on the quality of service delivery: a pre and post assessment of the healthcare consumer's experiences in Ghana. Electron J Inform Syst Dev Countries 2024; 90: e12318.

38.

Ndayishimiye

Lopes

Middleton

. A systematic scoping review of digital health technologies during COVID-19: a new normal in primary health care delivery. Health Technol (Berl) 2023; 13: 273–284.

39.

Erku

Khatri

Endalamaw

, et al. Digital health interventions to improve access to and quality of primary health care services: a scoping review. Int J Environ Res Public Health 2023; 20: 6854.

40.

McDonald

Foley

Verheij

, et al. Data to knowledge to improvement: creating the learning health system. Br Med J 2024; 384: e076175.

41.

Ong

Hodgson

Small

, et al. Implementing a digital patient feedback system: an analysis using normalisation process theory. BMC Health Serv Res 2020; 20: 1–16.

42.

Jia

. Impact of digital infrastructure construction on the migrants’ utilization of basic public health services in China. BMC Health Serv Res 2024; 24: 61.

43.

Wong

Mavondo

Fisher

. Patient feedback to improve quality of patient-centred care in public hospitals: a systematic review of the evidence. BMC Health Serv Res 2020; 20: 30.

44.

Ferreira

Vieira

Pedro

, et al. Patient satisfaction with healthcare services and the techniques used for its assessment: a systematic literature review and a bibliometric analysis. Healthcare 2023; 11: 639.

45.

Reeves

West

Barron

. Facilitated patient experience feedback can improve nursing care: a pilot study for a phase III cluster randomised controlled trial. BMC Health Serv Res 2013; 13: 59.

46.

Al-Jabr

Twigg

Scott

, et al. Patient feedback questionnaires to enhance consultation skills of healthcare professionals: a systematic review. Patient Educ Couns 2018; 101: 1538–1548.

47.

Woods

Martin

Khor

, et al. The right care in the right place: a scoping review of digital health education and training for rural healthcare workers. BMC Health Serv Res 2024; 24: 1011.

48.

Abell

Naicker

Rodwell

, et al. Identifying barriers and facilitators to successful implementation of computerized clinical decision support systems in hospitals: a NASSS framework-informed scoping review. Implement Sci 2023; 18: 32.

49.

Sambo

BNMN

. Can training of health care workers improve data management practice in health management information systems: a case study of primary health care facilities in Kaduna State, Nigeria. Afr Med J 2018; 30: 289.

50.

Shortreed

Cook

Coley

, et al. Challenges and opportunities for using big health care data to advance medical science and public health. Am J Epidemiol 2019; 188: 851–861.

51.

Malhan

Sadeghi

Pavur

, et al. Healthcare information management and operational cost performance: empirical evidence. Eur J Health Econ 2024; 25: 963–977.

52.

Siyam

York

, et al. The burden of recording and reporting health data in primary health care facilities in five low- and lower-middle income countries. BMC Health Serv Res 2021; 21: 91.

53.

George

Jack

Gauld

, et al. Impact of health system governance on healthcare quality in low-income and middle-income countries: a scoping review. BMJ Open 2023; 13: e073669.

54.

Neve

Fyfe

Hayhoe

, et al. Digital health in primary care: risks and recommendations. Br J Gen Pract 2020; 70: 609–610.

55.

Reeves

Coulter

Jenkinson

, et al. Development and pilot testing of questionnaire for use in the acute NHS Trust inpatient survey programme. 2002.

56.

Staniszewska

Boardman

Gunn

, et al. The Warwick Patient Experiences Framework: patient-based evidence in clinical guidelines. Int J Qual Health Care 2014; 26: 151–157.

57.

Duffy

Christie

Moreno

. The challenges toward real-world implementation of digital health design approaches: narrative review. JMIR Hum Factors 2022; 9: e35693.

58.

Shandley

Hipp

Anderson-Bialis

, et al. Patient-centered care: factors associated with reporting a positive experience at United States fertility clinics. Fertil Steril 2020; 113: 797–810.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.20 MB