Barriers to neurodevelopmental care: Caregiver pathways, stigma, and delayed presentation

Study design and setting

This cross-sectional study was conducted at tertiary center in Uttar Pradesh between May 2021 and June 2024. The center serves as a referral hub for 11 districts in northwestern India, encompassing rural referrals from peripheral health facilities and semi-urban families from the surrounding catchment area.

Participants

Children aged 1–18 years with NDDs were eligible.

Inclusion criteria

Children with Diagnostic and Statistical Manual of Mental Disorders-5 (DSM-5) confirmed ADHD, ASD, or both, diagnosed through the multidisciplinary Child Development Clinic comprising consultant pediatricians, psychiatrists, and child psychologists.

Exclusion criteria

Children with profound sensory impairment, acute medical instability requiring intensive care, or whose caregivers declined participation.

We enrolled 110 children using consecutive sampling.

Data collection

Primary caregivers were interviewed face-to-face in Hindi/English using a semi-structured pro forma capturing sociodemographic characteristics, care-seeking pathways, and barriers.

The pro forma consisted of three sections:

1. Sociodemographic profile: Age, sex, residence, parental education, and socioeconomic status (Modified Kuppuswamy Scale).

2. Pathways to care: Chronology of help-seeking behaviors, including first point of contact (medical vs. non-medical) and duration between symptom recognition and tertiary consultation.

3. Barriers to care: A checklist of barriers derived from prior literature, including financial constraints, transportation difficulties, stigma, and cultural beliefs. Caregivers could report multiple barriers.^[2]

The instrument is in Supplementary Appendix 1. The instrument was developed after reviewing existing tools used in Indian NDD research,^[2,4] refined through expert review by two developmental pediatricians, and pilot-tested on 10 caregivers (excluded from the final sample). The present study reports sections pertaining to care-seeking pathways and barriers; barrier items were dichotomized as present versus absent for descriptive analysis. The instrument was study-specific and not subjected to formal psychometric validation, which is acknowledged as a limitation. As symptom recognition timing relied on retrospective caregiver recall, interviewers used developmental milestone anchoring and cross-referenced available medical records to improve temporal accuracy.

Operational definitions

Ethical considerations

The Institutional Ethics Committee of approved this study. Parents/guardians provided written consent. Children aged ≥7 years provided verbal assent when developmentally appropriate.

Statistical analysis

Data were analyzed in the Statistical Package for the Social Sciences 26.0. Categorical variables are presented as frequencies and percentages; continuous variables are presented as means ± standard deviation. Continuous variables were assessed for normality using the Shapiro– Wilk test. Chi-square tests (Fisher’s exact when expected counts <5) assessed associations between maternal education, residence, and delayed presentation (significance: p < 0.05). No a priori sample size calculation was performed; all eligible children presenting during the study period were enrolled consecutively. post hoc power analysis for the primary association (maternal education and delayed presentation, observed proportions 76.3% vs. 43.3%) confirmed adequate statistical power (>0.90) at α = 0.05. Missing data (<2%) were excluded. Variables significant in univariate analysis (p < 0.05), along with faith healer first contact as a pre-specified clinically relevant covariate, entered a binary logistic regression model. Model fit was assessed using Hosmer–Lemeshow goodness-of-fit; discriminative ability using receiver operating characteristic (ROC) analysis with area under the curve; predictive accuracy using classification accuracy and Nagelkerke R². Adjusted odds ratios (aOR) with 95% confidence intervals (CI) are reported.

Participant characteristics and pathways to care

Table 1 presents sociodemographic characteristics and care-seeking pathways. Half (51.8%) were aged 5–10 years at presentation; 21.8% before age 5. Males comprised 58.2% of the sample. Most families resided in rural areas (80.9%) and belonged to low socioeconomic strata (59.1%); 72.7% of mothers had ≤8 years of education.

Delayed presentation (>12 months between symptom recognition and tertiary consultation) occurred in 67.3% (n = 74). Initial help-seeking was predominantly non-medical: 90.0% (n = 99) first consulted faith healers or traditional practitioners.

Barriers to care

Caregivers reported multiple barriers [Table 2 and Figure 1]. Stigma-related avoidance was most common (90.9%), followed by preference for faith-based interventions (90.0%) and financial constraints (83.6%).

Transportation difficulties affected 77.3% and geographic distance >50 km affected 73.6%. Time constraints were reported by 59.1%, and caregiver availability issues by 45.5%. In addition, 40.0% cited COVID-19 pandemic disruptions as causing a lack of accessible services. The belief that the child would improve without treatment was reported by 38.2%. Language barriers were least common (10.0%), primarily among migrant families.

Close

Figure 1: Caregiver-reported barriers.

Export to PPT

Stigma subtypes and psychological barriers

Analysis of individual stigma items (Section C) revealed that 67 caregivers (60.9%) endorsed concerns about the adverse impact on marriage prospects of the affected child or healthy siblings. Fear of community gossip was reported by 45 (40.9%) and fear of school rejection by 17 (15.5%). Caregiver acceptance was assessed using a single item (Section E) rated on a 4-point scale from complete denial to full acceptance; at initial assessment, 62 caregivers (56.4%) scored in the denial or minimal acceptance categories.

Factors associated with delayed presentation

Table 3 shows associations between sociodemographic factors and delayed presentation. Children of mothers with ≤8 years of education were more likely to present late (76.3% vs. 43.3%; χ² = 11.2, p = 0.001).

Delayed presentation occurred in 85.7% of urban/semi-urban families versus 62.9% of rural families (χ² = 4.0, p = 0.045). Possible explanations for this finding are explored in the discussion.

Multivariate analysis: Independent predictors of delayed presentation

To identify independent predictors while controlling for confounding, variables significant in univariate analysis (maternal education, residence, socioeconomic status, distance from center, and first contact type) were entered into a binary logistic regression model with delayed presentation as the outcome [Table 4]. Figure 2 illustrates the relative strength of independent predictors of delayed presentation derived from the multivariable logistic regression model. The model demonstrated excellent fit (Hosmer-Lemeshow χ² = 3.66, p = 0.600), explained 39.6% of variance in the outcome (Nagelkerke R² = 0.396), and correctly classified 80.9% of cases. Discriminative ability was excellent, with an area under the ROC curve of 0.82 (95% CI: 0.74–0.90), indicating strong ability to distinguish families with delayed versus timely presentation [Figure 3].

Close

Figure 2: Relative strength of independent predictors of delayed presentation. CI: Confidence interval; SES: Socioeconomic status; OR: Odds ratio.

Export to PPT

Close

Figure 3: Receiver operating characteristic curve (ROC) for the multivariable logistic regression model.

Export to PPT

Low socioeconomic status emerged as the strongest independent predictor, with families from economically disadvantaged backgrounds having nearly 19-fold higher odds of presenting beyond 12 months compared to middle/high socioeconomic status families (aOR: 18.67, 95% CI: 4.47– 77.90, p < 0.001). This effect remained highly significant even after adjusting for maternal education, residence, and distance.

Distance >50 km from the tertiary center was the second strongest independent predictor (aOR: 4.47, 95% CI: 1.57– 12.68, p = 0.005), indicating that geographic barriers operate independently of economic status. Maternal education ≤8 years retained marginal independent significance (aOR: 2.99, 95% CI: 1.04–8.56, p = 0.041), though its effect was substantially weaker than that of socioeconomic status.

Importantly, two variables that showed significant bivariate associations lost significance in multivariate analysis. Urban/semi-urban residence did not independently predict delayed presentation (aOR: 0.68, 95% CI: 0.24–1.95, p = 0.477), indicating that the apparent “urban delay” observed in bivariate analysis was confounded by other factors, likely referral pathway differences. Similarly, consulting a faith healer as first contact did not independently affect presentation timing (aOR: 1.07, 95% CI: 0.22–5.14, p = 0.933), suggesting that families pursue spiritual and medical pathways in parallel rather than sequentially.

The dominance of non-medical pathways

Strikingly, 90.0% first consulted faith healers or traditional practitioners. This far exceeds the 40–60% reported in other Indian tertiary center studies.^[4] Previous research attributed this to poor awareness or cultural beliefs; our data suggest possible structural displacement, though this interpretation warrants confirmation through qualitative methods.^[2] With 80.9% from rural areas lacking specialists and early intervention services, traditional healers serve as de facto primary care rather than chosen alternatives. However, faith healer consultation did not independently predict delayed presentation after adjusting for socioeconomic status, distance, and maternal education, suggesting it reflects structural displacement and medical pluralism rather than causing diagnostic delay.

Low maternal education entrenches this reliance: 72.7% of mothers had ≤8 years of schooling. Maternal education was significantly associated with delayed presentation (76.3% delay in less-educated vs. 43.3% in more-educated families; χ² = 11.2, p = 0.001), corroborating that maternal education proxies health literacy, decision-making autonomy, and information access. Without medical frameworks, families default to cultural illness models, necessitating spiritual intervention.^[2]

Stigma as a social and economic barrier

Stigma-related avoidance was the most pervasive barrier (90.9%), considerably exceeding the 45–60% in broader Indian psychiatric/disability literature.^[6] Notably, 60.9% feared diagnosis would jeopardize marriage prospects of both the affected child and healthy siblings – “affiliate stigma” or “courtesy stigma.”^[7] We use “marriage penalty” to describe this specific subset of affiliate stigma. While overlapping with broader stigma constructs, 60.9% explicitly endorsed marriage-related concerns when directly queried, distinguishing this from general community gossip (40.9%) or school-related fears (15.5%). These categories are not mutually exclusive.

This stigma compels families to conceal the condition, delaying diagnosis until symptoms become socially or educationally unmanageable. Unlike Western contexts where stigma relates to individual concerns, our findings echo collectivist societies where neurodevelopmental diagnosis is viewed as familial “genetic taint” threatening family social capital and marriageability.^[8] In patriarchal, endogamous communities with arranged marriages, disability disclosure can eliminate marriage opportunities for all children, creating incentives for secrecy.^[9]

In addition, 40.9% feared community gossip and 15.5% feared school rejection. These social risks translate to economic vulnerabilities where community reputation determines access to credit, employment, and support networks. Stigma functions not merely as a psychological barrier but also as a rational calculation of social and economic risk.

The “severity filter” and gender disparity

Our male-to-female ratio of 1.4:1 (58.2% male) deviates significantly from the expected 3:1–4:1 in ASD/ADHD clinic populations globally.^[10] Rather than indicating a higher prevalence of NDD in females, this likely reflects a “severity filter” inherent to tertiary referral centers in low-resource contexts.

Given that 73.6% traveled >50 km and 83.6% faced financial barriers, families likely prioritize costly tertiary care for females only when symptoms are profound – severe intellectual disability or complete speech absence.^[11] Milder presentations in girls, particularly subtle internalizing symptoms in high-functioning ASD, may remain undiagnosed due to lower social disruption and financial burden.^[12]

Conversely, boys with externalizing behaviors (hyperactivity, aggression, disruptive conduct) may present earlier due to higher social visibility and cultural intolerance of male behavioral deviance. This differential referral threshold artificially compresses gender ratios at tertiary centers, masking true population prevalence and leaving undiagnosed girls in communities. This hypothesis warrants further investigation through population-based screening studies. This gender-differential threshold was reflected in caregiver expectations. When asked about desired functional outcomes, caregivers of girls frequently expressed modest domestic goals: “Ghar ka kaam seekh le, bas itna kaafi hai” (“If she learns housework, that’s enough”). In contrast, caregivers of boys articulated economic participation as minimal acceptable competence: “Dukaan par baith sake, paison ka hisaab kar le” (“If he can sit at the shop and handle money”). These gendered functional thresholds may partially explain why girls with preserved self-care skills but significant cognitive impairment remain under-referred, while boys with comparable abilities but disruptive behaviors prompt tertiary consultation.

Structural constraints: Distance, infrastructure, and opportunity cost

Although 83.6% reported financial constraints, only 20.9% cited lost wages as the primary burden. This discrepancy reflects economic barriers in agricultural and informal-sector economies. Many families engaged in subsistence farming or self-employment rather than wage labor, meaning the primary barrier was catastrophic out-of-pocket expenditure for travel, accommodation, and therapy rather than missed workdays.^[13]

With 80.9% reporting a complete absence of local services (speech therapy, occupational therapy, special education), the “distance penalty” acts as a hard exclusion. Transportation difficulties affected 77.3%, and for those >50 km from the center, repeated therapy visits became financially and logistically prohibitive.^[14]

In addition, 40.0% cited COVID-19 pandemic disruptions, reflecting fragile early intervention systems. The pandemic suspended already-sparse services and diverted healthcare resources from chronic developmental conditions, compounding pre-existing structural inequities.^[15] Distance independently predicted delayed presentation in multivariate analysis, confirming a “distance penalty” beyond socioeconomic disadvantage.

The urban-rural paradox: Pathways trump proximity

Although bivariate analysis showed higher delay rates in urban/semi-urban families (χ² = 4.0, ρ = 0.045), this did not persist after multivariate adjustment (aOR: 0.68; ρ = 0.477), indicating confounding. This inverts the typical “urban advantage,” showing structured referral pathways reduce diagnostic delay more effectively than geographic proximity.^[16]

Many rural children were identified through high-risk neonatal intensive care unit (NICU) follow-up programs enabling protocol-driven tertiary referral.^[17] Conversely, urban families often presented late after consulting multiple non-specialist providers lacking NDD expertise.^[18] This distinction underscores prioritizing structured referral systems and primary care training over merely increasing tertiary facility density.^[19,20]

Psychological barriers and the denial-delay cycle

The delay mechanism is not purely structural. We found 56.4% of caregivers in denial or demonstrating minimal acceptance at initial assessment. This psychological barrier acts synergistically with the belief that the child will “outgrow” the problem (38.2%), creating a “wait-andsee” loop extending beyond the critical 12-month early intervention window.^[21]

This aligns with stage models of parental adaptation describing initial denial as normative but potentially prolonged before active treatment-seeking occurs. Where neurodevelopmental disabilities lack cultural legitimacy – viewed as karmic consequences, spiritual afflictions, or transient variations – this denial phase may be extended.

Recent Indian parent-mediated intervention research emphasizes addressing parental acceptance and mental health as prerequisites to clinical adherence and efficacy.^[22,23]Our data are consistent with this: Even when financial and geographic barriers are overcome, psychological non-acceptance prevents sustained engagement. Interventions combining psychoeducation, peer support, and acceptance-building may complement service-delivery approaches.^[24]

Study limitations

Several limitations exist. First, sampling from a single tertiary center introduces selection bias toward families who overcame barriers to reach specialist care. True population burden of undiagnosed NDD among marginalized families remains unmeasured. Second, retrospective barrier data introduce recall bias regarding symptom recognition and care-seeking timelines. Third, the cross-sectional design precludes assessing how barriers evolve or how early interventions modify subsequent access. Fourth, a small urban subsample (n = 21) limits statistical power and generalizability of urban-rural comparisons. Certain variables of potential relevance – including parental age, total number of siblings, single-parent status, and paternal education – were not captured and represent unmeasured confounders. Finally, the absence of qualitative methods (in-depth interviews, focus groups) limits exploration of the mechanisms underlying caregiver decision-making and stigma experiences; a mixed-methods design is recommended for future studies.