Outcomes of congenital heart disease procedures in Pakistan: a retrospective cohort study using a publicly funded congenital heart surgery programme registry Authors: Mahim Malik, Asad Saulat Fatimi, Salman Shah, Muhammad Asim Khan, Faiz Rasool, Tariq Waqar, Usman Riaz, Zaigham Rasool Khalid, Syed Najam Hyder, Sheikh Abdul Malik, Muhammad Younas, Abdul Razzaq Mughal, Asif Ur Rehman, Farqad Alamgir, Adnan Khan, Masood Sadiq Published: 2026 ABSTRACT AbstractObjectivesThe objective of this study was to evaluate factors associated with in-hospital periprocedural mortality and length of stay (LOS) among patients treated through a publicly funded programme such as Pakistan’s Chief Minister’s Congenital Heart Surgery Program (CMCHSP), including impact of socioeconomic factors.Study designRetrospective cohort study using the CMCHSP registry.ParticipantsPatients undergoing congenital cardiac interventions between January 2024 and August 2025, across multiple public and private hospitals in Punjab, Pakistan.MethodsPrimary outcomes were in-hospital periprocedural mortality and hospital LOS. Procedural complexity was classified using the Risk Adjustment for Congenital Heart Surgery (RACHS) system. Socioeconomic status was assessed using Proxy Means Test scores. Multivariable logistic and linear regression models were used to evaluate risk-adjusted associations with mortality and LOS, accounting for clustering at the patient level.ResultsA total of 5431 patients underwent 5744 procedures, of which 55.3% were surgical and 44.7% catheter-based. The cohort was predominantly paediatric (87.1%) and had a median age of 6.0 years. Mortality increased significantly with procedural complexity, ranging from 1.46% in RACHS category 1 to 30.23% in category 4 (p<0.001). On multivariable analysis, each increase in RACHS category was associated with higher odds of mortality (adjusted OR 2.08, 95% CI 1.57 to 2.76). Urgent procedures were also associated with increased mortality (aOR 1.68, p=0.046). Socioeconomic status was not independently associated with mortality. Higher procedural complexity, treatment in public hospitals and greater distance-to-care were associated with prolonged LOS.ConclusionIn this large, programme-based cohort, outcomes were influenced by procedural complexity and clinical factors rather than socioeconomic status once access to care was achieved. These findings suggest that publicly financed congenital heart disease programmes can deliver equitable short-term outcomes, though system-level barriers continue to influence resource utilisation. STRENGTHS AND LIMITATIONS OF THIS STUDY Population-level assessment of the geographic and socioeconomic distribution of children registered for congenital heart disease surgery or catheter-based intervention through a publicly funded programme in Pakistan. Population-level assessment of the geographic and socioeconomic distribution of children registered for congenital heart disease surgery or catheter-based intervention through a publicly funded programme in Pakistan. Comprehensive evaluation of socioeconomic status via data from an independent national-level database that utilises a standardised tool (Proxy Means Test). Comprehensive evaluation of socioeconomic status via data from an independent national-level database that utilises a standardised tool (Proxy Means Test). We were unable to obtain data on socioeconomic status for all patients; therefore, an exploratory analysis was performed. We were unable to obtain data on socioeconomic status for all patients; therefore, an exploratory analysis was performed. Limited data availability for clinical outcomes beyond length of stay and procedural mortality. Limited data availability for clinical outcomes beyond length of stay and procedural mortality. Introduction Congenital heart disease (CHD) is the most common congenital anomaly worldwide and a leading cause of infant and childhood mortality from non-communicable diseases.1 2 Advances in surgery and interventional cardiology over the past four decades have transformed CHD from a largely fatal condition into a chronic disease compatible with long-term survival in high-income countries.3 However, these gains have been unevenly distributed. In low and middle-income countries (LMICs), limited diagnostic capacity, delayed referral, financial barriers and shortages of specialised centres continue to restrict access to timely definitive care, resulting in substantial preventable morbidity and mortality.4 5 Pakistan bears a substantial CHD burden given its large birth cohort, yet population-level data on outcomes of congenital cardiac interventions remain scarce. Care delivery is highly centralised, with advanced surgical and catheter-based services concentrated in a small number of tertiary institutions.1 2 Publicly funded initiatives such as the Chief Minister’s Congenital Heart Disease Programme (CMCHSP) aim to expand access for underprivileged populations by covering the cost of treatment at designated centres.6 While such programmes may reduce financial barriers, it remains unclear whether socioeconomic disadvantage continues to influence perioperative outcomes once patients enter the treatment pathway. Procedural complexity is already well-established as a dominant determinant of perioperative outcomes in congenital heart surgery.7 8 In resource-constrained settings, however, outcomes may also be shaped by factors that determine when and how patients enter the treatment pathway, including socioeconomic disadvantage, geographic access, referral delays and institutional capacity. Evidence from other surgical fields suggests that social determinants of health, including poverty, education and geographic isolation, can influence outcomes through delayed presentation, poorer baseline health, limited postoperative support and reduced access to follow-up care.9 10 Although publicly reimbursed programmes may reduce out-of-pocket expenditure, they may not fully address upstream delays in diagnosis and referral, regional variation in service capacity, or downstream differences in perioperative care and resource utilisation. This study analyses a large registry of patients undergoing congenital cardiac procedures within a publicly funded programme in Pakistan. The primary aim was to evaluate clinical and health-system factors associated with in-hospital periprocedural mortality and length of stay (LoS) among patients treated through CMCHSP, including procedural complexity, urgency, hospital sector and distance-to-care. We also aimed to assess whether household socioeconomic status, measured using the Proxy Means Test (PMT) score where available, was associated with short-term outcomes after entry into the programme. By examining CHD interventions in Pakistan, this study aimed to provide evidence on whether publicly financed specialised care can achieve equitable outcomes across socioeconomic strata, while identifying areas for further improvement in CHD service delivery in Pakistan and other similar LMIC settings. Methods Study design and data sources This was a retrospective observational cohort study using data from the Chief Minister’s Congenital Heart Surgery Programme (CMCHSP) registry from Pakistan, encompassing all patients who were diagnosed with CHD and registered for surgery or interventional cardiac catheterisation between January 2024 and August 2025. The CMCHSP registry captures demographic characteristics, diagnoses, procedural details, hospital information, socioeconomic indicators and clinical outcomes for patients evaluated and treated under the programme. Overview of the CMCHSP programme The CMCHSP is a government-funded initiative in the province of Punjab designed to improve access to timely and equitable care for CHD patients in Pakistan. The programme aims to address financial and capacity barriers by reimbursing the full package cost of cardiac interventions and surgeries for eligible patients. This is particularly relevant in Pakistan, where household out-of-pocket payments remain a major source of healthcare financing, accounting for nearly half of total health expenditure in the National Health Accounts 2021–2022 and 52.9% of current health expenditure in World Bank estimates for 2023, while private insurance and social security mechanisms contribute only marginally.11 12 The health system in Punjab comprises government-funded public hospitals alongside fee-for-service private hospitals; however, comprehensive congenital cardiac services in the public sector are concentrated in a small number of tertiary centres with limited specialist capacity and substantial procedural demand. A key feature of CMCHSP is its public–private partnership model, whereby patients are initially registered through public hospitals and may be referred to empanelled private-sector institutions once public-sector capacity is fully utilised. This approach is intended to expand surgical capacity, reduces wait times and enables access to specialised care that may not be uniformly available in the public sector alone. To ensure transparency, cost control, and uniformity in service delivery, the programme has established standardised package reimbursement rates for a wide range of cardiology procedures and cardiac surgeries across participating hospitals (online supplemental table 1). The programme has also allocated resources to strengthen public-sector capacity, including procurement of durable equipment and recruitment of nursing and ancillary staff. All participating centres are currently located in Punjab; however, although CMCHSP is an initiative of the Government of Punjab, patients from other parts of Pakistan where tertiary-level care is unavailable or unaffordable may also receive treatment through the programme. Public-sector programme capacity and procedural volume through August 2025 are summarised in online supplemental table 2, and the CMCHSP has been summarised in figure 1.13 A summary of the access pathway for and barriers being addressed by the Chief Minister’s Congenital Heart Surgery Programme (CMCHSP). CHD, congenital heart disease. Study population All adult and paediatric patients with a confirmed CHD diagnosis recorded in the CMCHSP registry were eligible for inclusion. All participating centres in the CMCHSP registry have been documented in online supplemental table 3. Patient and public involvement Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research. Data acquisition and variable computation Pakistan district centroid coordinates for distance-to-care data were obtained from the United Nations Office for the Coordination of Humanitarian Affairs (UN-OCHA) Humanitarian Data Exchange (HDX).14 Hospital coordinates were obtained through publicly available information from hospital websites. Distance-to-care was calculated as the geographic (great-circle) distance in kilometres between the centroid of the patient’s home district and the centroid of the treating hospital. Where both CMCHSP and National Socioeconomic Survey (NSER) district data were available, NSER district was prioritised to ensure geographic consistency. CHD severity was classified according to the American College of Cardiology/American Heart Association (ACC/AHA) adult CHD anatomic complexity framework (2018) and mapped to paediatric diagnoses where needed.15 Procedural complexity for surgical interventions was classified according to the Risk Adjustment for Congenital Heart Surgery (RACHS)−1 scoring system.16 Patients with rheumatic or non-congenital valvular disease and patients without a recorded diagnosis were retained in descriptive summaries to reflect overall programme workload, but these categories were not assigned congenital anatomic complexity or RACHS-1 classifications. Accordingly, they were not included in RACHS-based analyses. Details regarding assignment of RACHS scores to procedure names listed in the CMCHSP registry are included in online supplemental table 4. To assess socioeconomic status, CMCHSP data were linked to the NSER database, which provides PMT scores, a nationally standardised poverty metric derived from household-level socioeconomic indicators. Scores range from 0 to 100, with lower PMT scores indicating greater socioeconomic deprivation. A score of 32 or less is used to identify those who would qualify for social welfare programmes.17 Outcome measures The primary outcomes of interest were in-hospital periprocedural mortality and hospital LOS. In-hospital periprocedural mortality was defined as death occurring during the same hospitalisation as the procedure, based on discharge status recorded in the registry. The registry did not systematically capture deaths occurring after discharge, and therefore 30-day operative mortality could not be assessed. LOS was calculated as the number of days from admission to discharge. LOS analyses were restricted to patients who underwent a procedure. Data analysis Data cleaning Duplicate records were removed using a composite key of medical record number, guardian national identity card (NIC) number, patient name, time of admission and date of birth. These identifiers were used solely for data cleaning and were not retained in the analytical dataset or included in any reporting. Multiple separate surgical or interventional hospitalisations of the same patient were treated as independent episodes of care. This was done because children with CHD frequently undergo staged evaluation and treatment, including diagnostic catheterisation, medical stabilisation and definitive repair across separate admissions. We also chose not to use the index admission as the unit of analysis because this would risk misclassifying diagnostic encounters as therapeutic events and excluding subsequent definitive interventions, thereby underestimating operative activity and biasing outcome estimates. Therefore, inferential analyses were conducted at the procedure level rather than the patient level, while descriptive statistics were reported at the patient level. Descriptive and inferential statistics All analyses were conducted in Python (V.3.11) using pandas and numpy for data management, scipy for hypothesis testing and correlations, and statsmodels for regression modelling. Continuous variables were assessed for distributional shape visually using Q-Q plots and histograms. Given that all continuous variables were not normally distributed, continuous data were summarised as medians and interquartile ranges (IQR). Categorical variables were summarised as frequencies and percentages. Group comparisons were performed using the Wilcoxon rank-sum test (or Kruskal-Wallis for >2 groups) for continuous variables and the χ² test for categorical variables. Mortality differences across ordered RACHS levels were evaluated using the Cochrane-Armitage test for trend. PMT scores were summarised across quartiles for descriptive reporting. Regression analysis Risk-adjusted associations with mortality were evaluated using multivariable logistic regression, and associations with LOS were evaluated using multivariable linear regression on log-transformed LOS. Due to right-skewed distribution, LOS was log-transformed using ln(LOS+1) prior to regression modelling. Results were back-transformed and reported as percent change in LOS. LOS analyses were only performed for survivors. Prespecified covariates included age at procedure, sex, PMT score, hospital type (public vs. private), procedure urgency (elective, semi-elective, urgent, emergency), and procedural complexity using the RACHS-1 category for surgical procedures. RACHS categories were treated as ordinal variables reflecting increasing procedural complexity. No automated variable selection was performed; all covariates were retained based on clinical and programmatic relevance. Admissions lacking these variables were excluded from multivariable analyses. Interventional cardiology procedures were included in descriptive analyses but excluded from primary risk-adjusted models because there was no comparable standardised risk classification. Because individual patients could contribute multiple admissions, standard errors (SEs) were adjusted for within-patient clustering using robust sandwich estimators clustered by unique patient identifiers (guardian NIC number, date of birth, and patient name). This approach accounts for non-independence of observations while preserving all eligible admissions in the analysis. Therefore, model estimates should be interpreted as procedure-level associations. Results are reported as adjusted ORs (aORs) with 95% CIs for mortality and as per cent change in LOS with 95% CIs for LOS models. All analyses were two-sided, and statistical significance was defined as p<0.05. Results Patient characteristics A total of 5431 unique patients with congenital or structural heart disease were included in the operated cohort, who underwent 5744 operations. 4733 (87.1%) were children aged ≤18 years and 697 (12.8%) were adults. The cohort showed a slight male predominance overall (53.2%). The median age at presentation was 6.00 years (IQR: 2.55–11.84). Among paediatric patients, the largest age group was 1–4 years (36.6%), followed by 5–9 years (30.2%) and infants <1 year (12.7%). Geographically, patients originated predominantly from Punjab (90.8%). In terms of disease severity based on the ACC/AHA classification, most cases were classified as moderate complexity (89.1%), while 7.1% had great (complex) CHD. In addition, most referrals were elective (71.8%) or semi-elective (19.5%). Socioeconomic information derived from PMT scores was available for 3279 patients (60.4%). The median PMT score was 32.95 (IQR: 27.89–39.22) overall, with similar distributions among paediatric and adult patients (p=0.08). Among patients with available data, 1506 (27.7%) fell below the PMT threshold of 32, indicating eligibility for government financial support; this proportion was statistically significantly higher among children (28.2%) than adults (24.8%; p=0.03). Besides this, the median distance-to-care in the overall cohort was 84.6 km (IQR: 29.1–143.0). Additional cohort characteristics are present in table 1. Table 1Demographic and referral characteristics of patients (N=5431)CharacteristicAll patientsChildren (≤18 years)Adults (>18 years)Total patients5431 (100%)4733 (87.1%)697 (12.8%)Gender Male2889 (53.2%)2612 (55.2%)277 (39.7%) Female2541 (46.8%)2120 (44.8%)420 (60.3%) Transgender1 (<0.1%)1 (<0.1%)0 Age, years—median (IQR)6.00 (2.55–11.84)5.05 (2.17–9.00)26.01 (20.96–35.94)Age brackets (years) <1602 (11.1%)602 (12.7%)– 1–41733 (31.9%)1733 (36.6%)– 5–91431 (26.3%)1431 (30.2%)– 10–14720 (13.3%)720 (15.2%)– 15–18247 (4.5%)247 (5.2%)– >18 (adults)697 (12.8%)–697 (100%)Province Punjab4934 (90.8%)4299 (90.8%)634 (91.0%) Khyber Pakhtunkhwa321 (5.9%)282 (6.0%)39 (5.6%) Sindh9 (0.2%)6 (0.1%)3 (0.4%) Balochistan21 (0.4%)19 (0.4%)2 (0.3%) Azad Jammu & Kashmir132 (2.4%)114 (2.4%)18 (2.6%) Gilgit-Baltistan12 (0.2%)11 (0.2%)1 (0.1%)CHD severity (ACC/AHA classification) Simple0 (0.0%)0 (0.0%)0 (0.0%) Moderate4839 (89.1%)4188 (88.5%)651 (93.4%) Great (complex)383 (7.1%)369 (7.8%)14 (2.0%) Rheumatic and non-congenital valvular disease48 (0.9%)46 (1.0%)2 (0.3%) No diagnosis given161 (3.0%)131 (2.8%)30 (4.3%)Urgency category Elective3900 (71.8%)3361 (71.0%)539 (77.3%) Semi-elective1058 (19.5%)931 (19.7%)127 (18.2%) Urgent388 (7.1%)366 (7.7%)21 (3.0%) Emergent34 (0.6%)31 (0.7%)3 (0.4%) Not recorded51 (0.9%)44 (0.9%)7 (1.0%)Socioeconomic data Patients with PMT data3279 (60.4%)2868 (60.6%)411 (59.0%) Median PMT score32.95 (27.89–39.22)32.80 (27.83–39.20)33.68 (28.28–39.42) On government support (PMT<32)1506 (27.7%)1333 (28.2%)173 (24.8%)Hospital sector Private1577 (29.0%)1412 (29.8%)165 (23.7%) Public3854 (71.0%)3321 (70.2%)532 (76.3%) Distance-to-care (km)—median (IQR)84.6 (29.1–143.0)84.7 (29.1–146.7)82.2 (28.5–129.9)ACC/AHA, American College of Cardiology/American Heart Association; CHD, congenital heart disease; PMT, Proxy Means Test. Demographic and referral characteristics of patients (N=5431) ACC/AHA, American College of Cardiology/American Heart Association; CHD, congenital heart disease; PMT, Proxy Means Test. Diagnosis profile Tetralogy of Fallot (TOF) was the most frequently encountered diagnosis, accounting for 1427 cases, followed by ventricular septal defect (VSD; 1317), patent ductus arteriosus (PDA; 1163) and atrial septal defect (ASD; 1129). Together, these four lesions constituted the majority of presentations, reflecting the predominance of common acyanotic defects and TOF within the cohort. Other relatively frequent congenital lesions included pulmonary valve stenosis (398 cases) and transposition of the great arteries (TGA; 274 cases). Atrioventricular septal defect (155), congenital aortic valve stenosis (129) and coarctation of the aorta (113) were also commonly observed. Detailed diagnostic data for the cohort is available in online supplemental table 5. Procedure profile Surgical procedures Our cohort comprised 3175 (55.3%) surgical interventions. TOF repair was the most commonly performed surgical procedure, accounting for 891 operations. This was followed by VSD closure (743 cases) and ASD closure (719 cases) and PDA ligation (193 cases). In addition, cavopulmonary shunts, including Glenn or hemi-Fontan procedures, were performed in 75 patients, while pulmonary artery banding was undertaken in 71 cases. Interventional procedures Our cohort comprised 2569 (44.7%) catheter-based procedures. Diagnostic cardiac catheterisation accounted for 1013 procedures and primarily contributed to diagnostic evaluation, preoperative assessment and postoperative surveillance rather than definitive treatment. Among therapeutic catheter-based interventions, PDA device closure was the most common (814 cases), followed by ASD device closure (212 cases), pulmonary valvuloplasty for pulmonary stenosis (157 cases) and balloon atrial septostomy for TGA (92 cases). Outcomes Overall mortality was 3.8% (n=221). Median LoS for the overall cohort was 3.54 days (IQR 1.51–9.18), with substantially longer hospitalisation following surgical procedures (7.51 days, IQR 4.45–17.45) compared with catheter-based interventions (1.48 days, IQR 0.55–2.27). Additional details regarding surgical and non-surgical interventions are available in online supplemental table 6A and 6B. Univariable association analyses Surgical complexity (RACHS Score) Mortality increased progressively with surgical complexity. Using the RACHS-1 classification, mortality rose from 1.46% in category 1 procedures to 30.23% in category 4 procedures (p for trend <0.001). Length of hospital stay also increased significantly with risk level (Kruskal-Wallis p<0.001). These data are presented in table 2. Table 2Risk-adjusted outcomes by RACHS-1 risk level (surgical procedures only)RACHS-1Procedures (n)Deaths (n)In-hospital mortality %LOS median (IQR), days1892131.464.62 (3.51–9.53)217761488.338.43 (4.58–20.19)3413389.2011.53 (6.44–25.64)4431330.2323.55 (10.56–46.97)Uncategorised5147.847.47 (2.61–19.37)LOS, length of stay; RACHS, Risk Adjustment for Congenital Heart Surgery. Risk-adjusted outcomes by RACHS-1 risk level (surgical procedures only) LOS, length of stay; RACHS, Risk Adjustment for Congenital Heart Surgery. Socioeconomic data In-hospital periprocedural mortality increased progressively across socioeconomic quartiles, from 2.42% in the poorest group to 4.38% in the richest group. Although this pattern suggested a graded association between higher PMT scores and mortality, differences across quartiles did not reach statistical significance (χ²=6.9, p=0.07). Median postoperative LoS was similar across quartiles (range 3.44–3.57 days), with no statistically significant difference observed (Kruskal-Wallis p>0.05). These data are present in table 3. Table 3Socioeconomic predictors (N=3471)PMT quartilesProcedures (n)Deaths (n)Mortality %LOS median (IQR), daysQ1 poorest (3.50–27.82)868212.423.52 (1.56–8.50)Q2 (27.82–32.94)868273.113.55 (1.51–8.50)Q3 (32.94–39.21)868323.693.57 (1.53–8.52)Q4 richest (39.21–75.66)867384.383.44 (1.47–9.47)LOS, length of stay; PMT, Proxy Means Test. Socioeconomic predictors (N=3471) LOS, length of stay; PMT, Proxy Means Test. Multivariable regression In-hospital periprocedural mortality On multivariable logistic regression analysis, procedural complexity was the strongest independent predictor of mortality. Each one-level increase in RACHS-1 category was associated with a more than twofold increase in odds of death (adjusted OR: 2.08 (95% CI 1.57 to 2.76), p<0.001). In addition, urgent procedures were more likely to result in periprocedural mortality as compared with elective procedures (adjusted OR: 1.68 (95% CI 1.01 to 2.80), p=0.046). Age, PMT score, hospital type, procedure urgency, distance-to-care and hospital type were not statistically significantly associated with mortality. Length of hospital stay Among survivors, procedural complexity was also associated with prolonged hospitalisation. Each increase in RACHS category corresponded to a 20.9% increase in LOS (95% CI 15.5 to 26.5%, p<0.001). Procedures performed in public hospitals were associated with markedly longer LOS compared with private institutions (+108%, 95% CI 95% to 121%, p<0.001). Semielective procedures were also statistically significantly associated with longer hospital stays relative to elective procedures (+78%, p<0.001). In addition, distance-to-care was statistically significantly associated with longer stays, with each 100 km increase in distance corresponding to a +4.6% increase in LOS (95% CI 2.1% to 7.3%, p<0.001). Age, sex and PMT score showed no statistically significant association with LOS. These results are presented in table 4. Table 4Risk-adjusted predictors of in-hospital periprocedural mortality and length of stay (LOS) among surgical proceduresVariableIn-hospital periprocedural mortality (adjusted OR, 95% CI)P valueLOS (% change, 95% CI)P valueAge (per year)0.98 (0.95 to 1.01)0.18−0.6% (−1.3% to 0.1%)0.09Female vs male1.12 (0.82 to 1.54)0.483.0% (−3.1% to 9.5%)0.33PMT score (per point)1.01 (0.98 to 1.03)0.55−0.1% (−0.33% to 0.16%)0.46Distance to care (per 100 km)1.06 (0.81 to 1.38)0.684.6% (2.1% to 7.3%)<0.001Public vs private hospital1.19 (0.78 to 1.81)0.41107% (95% to 121%)<0.001Urgency category Semi-elective vs elective1.22 (0.79 to 1.88)0.3676% (61% to 93%)<0.001 Urgent vs elective1.68 (1.01 to 2.80)0.04641% (14% to 74%)0.003 Emergency vs elective2.11 (0.94 to 4.74)0.0763% (17% to 127%)0.007 RACHS-1 category (per level increase)2.08 (1.57 to 2.76)<0.00120.9% (15.5% to 26.5%)<0.001Mortality model sample: 1880 admissions, 1859 patients; events: 113 deaths (6.0%) LOS model sample: 1767 admissions, 1750 patients; median LOS: 7.23 days.RACHS, Risk Adjustment for Congenital Heart Surgery. Risk-adjusted predictors of in-hospital periprocedural mortality and length of stay (LOS) among surgical procedures Mortality model sample: 1880 admissions, 1859 patients; events: 113 deaths (6.0%) LOS model sample: 1767 admissions, 1750 patients; median LOS: 7.23 days. RACHS, Risk Adjustment for Congenital Heart Surgery. Discussion This study provides one of the first large-scale evaluations of procedural outcomes within a publicly funded CHD programme in Pakistan and contributes to a limited but growing body of evidence on CHD care delivery in LMICs. Most existing literature focuses either on single-centre surgical outcomes or on access to care, whereas comprehensive programme-level analyses integrating socioeconomic, geographic and clinical factors are scarce.1 8 16 18 Consistent with the existing literature, we found that procedural complexity was the dominant determinant of perioperative outcomes, with mortality and LOS increasing steeply across RACHS risk categories.5 7 8 Furthermore, health-system factors, including urgency of intervention, hospital sector and geographic distance to care, were strongly associated with increased LOS and, therefore, resource utilisation. In addition, PMT score was not independently associated with in-hospital mortality among patients with available socioeconomic data; however, this finding should be interpreted cautiously because the analysis excludes patients who did not reach the programme and cannot assess unmet need, prereferral mortality or delays in diagnosis and referral. Together, these findings provide important insights into how large publicly funded CHD programmes function in resource-constrained settings. The operated cohort was predominantly paediatric (87%) with a median age of 6 years, reflecting delayed diagnosis and treatment compared with high-income countries, where many major CHD lesions are repaired in infancy. Similar age distributions have been reported in large CHD surgical series from India, China and sub-Saharan Africa, where late presentation remains common due to limited newborn screening, fragmented referral pathways and financial barriers.18–20 While the dominance of patients from Punjab mirrors the geographic location of programme implementation, over 10% patients travelled from other provinces for treatment. This highlights the positive spillover effects of such programmes in improving access for patients from other areas. The diagnostic spectrum, dominated by TOF, VSD, PDA and ASD, aligns with reports from national registries in India and Brazil where common, surgically correctable lesions account for the majority of operative volume.19 21 This phenomenon is widely documented in LMIC settings where early mortality, underdiagnosis or limited surgical capacity reduce survival to operative eligibility for more complex lesions.1 22 However, while the relatively lower representation of more complex procedures, such as neonatal surgeries and single ventricle physiology, suggests persistent attrition prior to definitive care, our experience shows that improving access also provides the opportunity for some of these children to undergo more complex procedures. Additionally, hospitals and care teams can review and audit outcomes and thereby potentially improve surgical capacity and outcomes. The substantial proportion of catheter-based procedures (45%), particularly diagnostic catheterisations, underscores the hybrid nature of modern CHD care. Similar programme evaluations from India’s publicly funded paediatric cardiac schemes have demonstrated that interventional cardiology capacity is a prerequisite for scaling surgical services, as catheterisation plays a central role in diagnosis, staging and postoperative management.23 24 Socioeconomic disadvantage is a well-established determinant of CHD outcomes globally, primarily through its impact on access to care. Studies from India’s Hridyam Programme have shown that publicly funded programmes can dramatically increase CHD surgical volumes among low-income families while narrowing disparities in survival.24 In the present cohort, more than one quarter of patients with available data qualified for government financial support, confirming that the programme serves a socioeconomically vulnerable population. However, PMT score was not independently associated with perioperative mortality after adjustment. Along these lines, previous studies have postulated that once treatment is obtained at specialised centres, operative survival depends more on anatomical complexity and institutional capacity than on household income.25 26 However, in our dataset, this observation is best understood through an ‘access filter’ effect, whereby only patients who successfully navigate structural, financial and geographic barriers are the ones who reach definitive surgical care. As a result, the treated cohort represents a selected population in whom downstream outcomes are primarily driven by anatomical and procedural risk rather than household socioeconomic status. This has important implications for interpretation: the absence of a socioeconomic gradient in surgical outcomes should not be interpreted as absence of inequity, but rather as evidence that disparities are concentrated upstream in the care pathway, prior to entry into specialised treatment systems. The observed trend towards higher mortality in wealthier quartiles likely reflects case-mix differences rather than a true protective effect of poverty. In LMIC settings, families with greater resources may be more likely to pursue care for complex or late-stage disease, whereas poorer families may never reach tertiary services.1 4 27 28 Additionally, those with more socioeconomic advantage may have access to better neonatal care, allowing patients with complex disease to survive to definitive surgical care. Procedural complexity emerged as the strongest predictor of mortality, with each increase in RACHS category more than doubling the odds of death. This gradient is remarkably consistent with data from high-income countries and validates the applicability of RACHS-based risk stratification in LMIC contexts.16 29 Urgent procedures were also independently associated with mortality, reinforcing the critical importance of timely diagnosis and referral. Delayed presentation leading to haemodynamic instability or organ dysfunction is a well-recognised contributor to poor outcomes in LMIC CHD cohorts.4 22 The comparatively high mortality observed among higher complexity procedures also indicates that improved access alone is unlikely to be sufficient to optimise outcomes. Alongside earlier diagnosis and referral, programmes such as CMCHSP require delivery-level quality improvement focused on perioperative standardisation, multidisciplinary case review, risk-adjusted outcome monitoring and capacity building for complex congenital cardiac surgery. Quality improvement frameworks developed for paediatric and congenital cardiac services in LMICs provide a practical structure for integrating these efforts into publicly funded programmes.30 Aside from this, LOS was strongly influenced by system-level factors. Public hospital procedures were associated with markedly longer LOS compared with private institutions. This pattern mirrors findings from publicly funded cardiac programmes in India and Brazil, where differences in discharge infrastructure, operating room availabilities and social circumstances contribute to prolonged admissions in public facilities.19 21 23 Distance-to-care also independently predicted longer hospitalisation, providing quantitative evidence of geographic barriers even after access to tertiary care is achieved. Patients travelling long distances may require additional stabilisation before discharge or prolonged inpatient observation due to limited access to follow-up services near home. The findings demonstrate that high-complexity congenital cardiac care can be delivered with outcomes primarily determined by clinical risk rather than socioeconomic status once patients reach specialised centres. These findings have important implications for health system design and policy in LMIC settings. While financial protection mechanisms such as CMCHSP appear effective in enabling equitable perioperative outcomes among treated patients, further improvements in population-level CHD outcomes will depend on addressing barriers earlier in the care continuum. Strengthening foetal, newborn and early childhood screening, including wider implementation of pulse oximetry and community-level diagnostic pathways, may reduce delays in presentation.31 32 In parallel, the development of structured referral networks and regional cardiac hubs, supported by tele-echocardiography and digital triage systems, could improve timely access to specialised care while reducing the burden of long-distance travel. Within the hospital setting, our findings identify modifiable drivers of resource utilisation. The markedly longer LoS observed in public sector institutions suggests opportunities for improving efficiency through standardised perioperative pathways, enhanced discharge planning and the development of step-down or transitional care facilities for patients travelling long distances. These interventions may increase surgical throughput and reduce costs without requiring expansion of operative capacity, thereby improving overall programme sustainability. Future directions Several priorities emerge for future research and programme development. Upstream equity analyses are needed to examine diagnostic delays, referral pathways and barriers to initial presentation in order to better characterise unmet need and the true population burden of CHD. Regionalisation of care, through strengthened secondary-level diagnostic services and telecardiology networks, may help reduce delays and travel burdens for patients and families. Programme effectiveness should also be assessed using long-term outcomes beyond discharge, including survival, reintervention rates, functional status and neurodevelopmental outcomes. A central priority is the establishment of a robust outcome-monitoring framework supported by a dedicated national database for benchmarking across centres, modelled on registries such as the Society of Thoracic Surgeons Congenital Heart Surgery Database, capturing standardised perioperative variables, case complexity, risk-adjusted outcomes and long-term follow-up. Such a system would enable performance comparisons, identification of best practices and continuous quality improvement, while broader initiatives, such as standardised perioperative protocols, could reduce variability in LoS and mortality.33 Limitations This study is bound by some limitations. Our study should be interpreted as a programme-level evaluation rather than a national assessment of CHD care in Pakistan. Most patients were from Punjab, and all participating centres were located in Punjab during the study period, limiting generalisability to provinces and regions without comparable access to tertiary congenital cardiac services. The registry also does not include patients from non-covered regions who were never referred, remained on waiting lists, died before reaching care or received treatment outside CMCHSP; therefore, we could not directly contrast outcomes between covered and non-covered populations or estimate the programme’s impact on national CHD mortality. Socioeconomic data were incomplete, and although PMT scores were available for 60.4% of patients, missingness may have introduced selection bias. Consequently, the socioeconomic analysis should be considered exploratory and limited to patients who entered the treatment pathway. In addition, analyses were conducted at the procedure level, allowing accurate characterisation of perioperative outcomes but limiting inference about patient-level risk. Although clustering by patient identifier was used, residual correlation may persist. Furthermore, given the purpose of the CMCHSP registry is not CHD outcomes research, clinical outcomes were restricted to short-term mortality and LOS, and important confounders such as nutritional status, comorbidities and anatomical detail were not available. Besides this, mortality was defined using discharge status and therefore reflects in-hospital periprocedural mortality rather than 30-day operative mortality. As a result, deaths occurring after discharge but within 30 days of the procedure may have been missed, potentially leading to underestimation of early mortality. Finally, the registry includes patients treated through CMCHSP and does not capture self-funded patients treated outside the programme; therefore, we could not compare CMCHSP reimbursement rates, case mix or outcomes with those of privately funded patients receiving care in the same institutions. Conclusions In this large cohort from a publicly funded CHD programme in Pakistan, perioperative outcomes were influenced by procedural complexity and clinical urgency, while socioeconomic status did not independently predict short-term mortality after access to care was achieved. Length of hospital stay was associated with geographical distance and whether procedure was performed in public or private hospital, which may be a reflection of persistent structural barriers. These findings suggest that while financial protection programmes can enable equitable survival among treated patients, further gains will depend on strengthening referral pathways, expanding regional capacity and improving efficiency of care delivery. Supplementary material