In low- and middle-income countries (LMICs), obtaining timely and precise data on the cause of death (CoD) is often a challenge due to the absence of standardised methods for collecting such information [1]. This data are vital for not only understanding the health status of a population but also for designing and evaluating public health policies and programmes. Healthcare resources are often limited in LMICs, and a clear understanding of the leading CoD can greatly assist in resource allocation and priority setting [2].
In Pakistan, like many LMICs, autopsy is often limited to medico-legal cases such as criminal deaths, suspicious or unexplained fatalities, accidents, deaths in custody, unidentified bodies, and suspected medical negligence. It is rarely performed, due to cultural beliefs, religious considerations, and resource constraints [3,4]. As a result, relying solely on autopsy data for understanding mortality patterns is not feasible. It is also estimated that around 61.2% (95% confidence interval (CI) = 48.3–73.1%) autopsies occur at home, outside of health care facilities, resulting in a significant proportion of unrecorded or unexamined mortality cases [5]. This presents a substantial challenge in accurately documenting mortality patterns and identifying underlying CoD, particularly in rural or remote areas with limited access to health care services [6]. To improve the availability of mortality data, there is a growing need to explore alternative methods for determining the CoD. A promising method of gathering this data in resource constrained settings like Pakistan is verbal autopsy (VA).
Verbal autopsies involve structured, detailed interviews with family members or caregivers of the deceased to gather information on the symptoms and circumstances leading up to death [7]. Data are then analysed using standardised algorithms or expert reviews to assign a probable CoD [8]. Verbal autopsy offers several advantages in contexts where autopsies are not feasible. First, it can be implemented in both health care settings and at the community level, allowing for a broader capture of mortality data, especially in regions where most deaths occur outside of medical facilities [9]. Additionally, the VA method is cost-effective, making it more accessible to LMICs where resources for health care and post-mortem examinations are limited [10]. Studies have also shown that when combined with medical certification and vital registration systems, VA can provide valuable insights into mortality patterns and help bridge the gap in mortality data [11,12].
Although VA is being utilised in Pakistan and other low- and middle-income settings, prior studies are largely restricted to specific age groups, such as under five years, or have relied predominantly on hospital-based data. This study represents one of the first large-scale, community-based implementations of VA encompassing both adult and paediatric populations during and post-pandemic. By adopting a community-centred surveillance framework, the study significantly contributes to a more comprehensive understanding of mortality patterns and enhances the accuracy of CoD attribution, particularly in regions with limited health care access and where conventional mortality data systems are insufficient or absent.
METHODS
The study was conducted in two distinct peri-urban catchment areas, Ali Akbar Shah, and Bhains Colony in Karachi, Pakistan. The study was approved by the ethics review committee of the Aga Khan University (2021-6445-19025). These catchment areas encompass a culturally and ethnically diverse demographic and have robust Health and Demographic Surveillance Systems in place – providing a structured foundation for our investigation [13]. To assess cause-specific mortality through VA, we conducted interviews on a representative sample of 25% of all deaths in the two catchment areas of our study. With an estimated 6000 total deaths annually across both sites (3000 per site), this approach involved conducting VA for approximately 1500 fatalities per year (750 for each site). This sample size was expected to provide sufficient and representative data to identify leading CoD, particularly for poorly documented or misreported causes, and offer valuable insights for public health interventions. Data was collected from 2020–2022.
Data collection
Training of interviewers: equipping for sensitivity and accuracy
Prior to data collection, interviewers underwent a rigorous three-day training workshop using the World Health Organization (WHO) Trainer Manual [14]. Training was facilitated by a WHO-Questionnaire master trainer and a senior social scientist and covered essential topics including data collection techniques, transcription methods, research ethics, communication skills and sensitivity training. With an understanding of the local cultural norms during the burial and bereavement period and the possible psychological stress for both participants and study staff involved in VA interviews, the interviewers were also trained to navigate this environment with care and sensitivity.
Death identification and VA data collection
We were notified of deaths through a WhatsApp group used for real-time reporting by demographic surveillance teams. Families were approached one week after the death occurred. This delay was necessary to give families an appropriate time to grieve – ensuring sensitivity to local cultural practices and values. Upon approach, the family was carefully briefed on the VA process. This included a comprehensive explanation of the overall purpose of the procedure – to discern the circumstances leading to the demise. A consent form detailing the VA procedure was provided to the family. Research staff ensured that family members fully understood the consent form and VA process and made clear that participation was voluntary. Consent was taken from all interviewees prior to beginning the interview. Upon consent, trained interviewers led a structured and detailed dialogue with family members or individuals close to the deceased. This comprised of a wide array of questions, probing into the signs and symptoms preceding the death, the individual’s medical history, and contextual details that could shed light on potential causes. When available, interviews with the mother of the deceased were preferred. Where the mother was unavailable or deceased, the father or another close relative was approached for the VA questions. In cases where a close relative was not available during the initial approach by the team, the team would make a second visit a week later. A third visit was made in the following week if a close relative could still not be reached. The case was closed after three attempts. Each case status was noted as case closed, refused, or migrated.
Verbal autopsy interviews: utilising WHO 2016 Questionnaire
The WHO 2016 Verbal Autopsy Questionnaire was used to conduct interviews with next of kin or caregivers [14]. This standardised questionnaire comprises both open-ended questions for detailed symptom accounts and closed-ended questions with filter questions tailored for specific diseases. The questionnaire, presented in the local language, ensured accurate responses by facilitating a deeper understanding of the circumstances surrounding each death. To address the differing symptomatology and diagnostic challenges among age groups, separate modules within the WHO VA tool were employed for neonates (0–28 days), children (29 days–14 years), and adults (15 years and above). For neonatal deaths, interviews prioritised mothers and focused on antenatal history, birth-related events, and early-life symptoms such as poor feeding, lethargy, breathing difficulties, and signs of infection. For child deaths, interviews explored symptoms like fever, diarrhoea, seizures, cough, and signs of malnutrition, with caregivers or close relatives serving as primary respondents. Adult deaths required more detailed histories of chronic and acute symptoms, including chest pain, weight loss, jaundice, or stroke-like features. In reproductive-age women, additional questions captured obstetric history to identify potential maternal causes. All VA interviews were conducted by trained field workers using culturally sensitive approaches. To address this, preliminary meetings were held with community elders, local health workers, and religious leaders prior to study initiation. Their feedback helped refine the approach to identifying bereaved families, improve timing of consent requests, and adapt culturally appropriate language in the consent script ensuring community values were respected throughout the data collection process.
Determination of the CoD
Cause-of-death determination involved careful analysis of the collected data. Physicians played a crucial role in this process. Initially, one physician reviewed the case. It was then reviewed by another physician for confirmation [15]. If there was a disagreement between the first two assessments, a third physician would provide their opinion. The final CoD was determined by comparing the two main causes identified. [16]. This approach was implemented to enhance reliability and transparency in the classification of CoD. In addition, a trained psychologist accompanied the physicians during VA visits, provided grief support to parents, caregivers, or relatives of the deceased during the same visit as the VA form administration, and conducted regular counselling sessions with the study team to help them cope with the emotional toll of frequent interactions with bereaved families. This approach also fostered community trust and offered a replicable model for emotionally sensitive fieldwork.
Once the CoD was determined, the findings were shared with the community or the deceased person’s family in a respectful and culturally sensitive manner. A designated member of the research team particularly the physician who has identified the CoD and someone familiar with the community, would approach the family to deliver the information. This was done in person, ensuring a private and respectful environment to allow the family time to process the information. The findings were explained in a straightforward but compassionate way, with the family given the opportunity to ask questions or seek further clarification. Throughout this process, cultural practices and privacy were prioritised by following appropriate protocols and maintaining confidentiality. The research team made sure to provide support as needed, recognising the emotional and sensitive nature of the discussion. Figure 1 illustrates the process of VA.
Figure 1. Death notification and verbal autopsy data collection process. VA – verbal autopsy.
Data analysis
Trained physicians reviewed and analysed the VA questionnaires completed via interviews at the field sites. The validity of using physician review to determine CoDs for both children and adults was established, showing reasonable sensitivity and specificity for selected CoD [17]. Descriptive statistics were calculated for demographics and risk factors associated with death, providing a comprehensive overview of the patterns within the graveyard. The ICD-10 code was utilised to identify the specific CoD [15].
RESULTS
Out of 1500 recorded deaths, 716 (47.7%) occurred in children under five years of age, while 728 (48.5%) were in adults above 18 years. Most of the deaths – around 769 (51.2%) – took place in hospitals, with 731 (48.7%) occurring at home. Ethnically, Bengalis comprised the largest group with 463 deaths (30.8%), followed by Urdu-speaking individuals with 395 (26.3%) and Sindhis with 348 deaths (23.2%) Among 785 individuals with reported education, 39.6% had no formal schooling, while among 754 with marital status data, 31.7% were married and 11.6% were widowed. (Table 1).
Table 1. Sociodemographic characteristics of deceased individuals (n = 1500)
| Variable | Gender | ||
|---|---|---|---|
| Male, n = 793 (%) | Female, n = 707 (%) | Total, n = 1500 (%) | |
| Age | |||
| <5 | 384 (48.4) | 332 (46.9) | 716 (47.7) |
| 5–18 | 026 (3.2) | 030 (4.2) | 056 (3.7) |
| >18 | 383 (48.3) | 345 (48.8) | 728 (48.5) |
| Place of death | |||
| Home | 386 (48.6) | 345 (48.8) | 731 (48.7) |
| Hospital | 406 (51.2) | 363 (51.3) | 769 (51.2) |
| Ethnicity | |||
| Bengali | 268 (33.8) | 195 (27.6) | 463 (30.8) |
| Urdu | 192 (24.2) | 203 (28.7) | 395 (26.3) |
| Sindhi | 167 (21.0) | 181 (25.6) | 348 (23.2) |
| Punjabi | 049 (6.2) | 038 (5.3) | 087 (5.8) |
| Pushto | 036 (4.5) | 027 (3.8) | 063 (4.2) |
| Hazare Wala | 014 (1.7) | 011 (1.5) | 015 (1.6) |
| Hindko | 09 (1.1) | 07 (0.9) | 016 (1.0) |
| Saraiki | 07 (0.8) | 08 (1.1) | 015 (1.0) |
| Others | 26 (3.2) | 021 (2.9) | 047 (3.1) |
| Highest level of schooling (n = 785) | |||
| No formal education | 276 (35.1) | 318 (40.5) | 594 (39.6) |
| Primary | 080 (10.1) | 029 (3.7) | 109 (7.2) |
| Secondary | 035 (4.4) | 022 (2.8) | 057 (3.8) |
| Higher secondary | 016 (2.0) | 08 (1.0) | 024 (1.6) |
| Do not know | 01 (0.1) | - | 01 (0.0) |
| Marital status (n = 754) | |||
| Married | 291 (38.6) | 185 (24.5) | 476 (31.7) |
| Widowed | 037 (4.9) | 138 (18.3) | 175 (11.6) |
| Single | 066 (8.7) | 033 (4.3) | 099 (6.6) |
| Divorced | 07 (0.9) | 02 (0.2) | 09 (0.6) |
| Refused to answer | 01 (0.13) | - | 01 (0.0) |
A mix of non-communicable (NCDs) and communicable diseases contributed to mortality in our population. Among NCDs, acute cardiac disease (n/N = 189/1500, 12.6%), liver cirrhosis (n/N = 105/1500, 7.0%), and stroke (n/N = 65/1500, 4.3%) were most prevalent, indicating a significant burden of chronic health conditions. Additionally, communicable diseases like diarrheal disease (n/N = 96/1500, 6.4%), pneumonia (n/N = 62/1500, 4.1%), and sepsis (n/N = 52/1500, 3.4%) were notable CoD. For children under five, neonatal causes like sepsis (n/N = 91/716, 12.7%) and perinatal asphyxia (n/N = 84/728, 11.5%) were leading contributors. In the 5–18 years category, external causes like road traffic crashes (n/N = 5/56, 8.9%) predominated, along with communicable diseases like diarrhoea and pneumonia (n/N = 4/56, 7.1%). For adults over 18 years, deaths related to NCDs were more common, such as acute cardiac disease (n/N = 182/728, 25.0%), liver cirrhosis (n/N = 96/728, 13.1%), and stroke (n/N = 64/728, 8.8%). External causes, including road traffic crashes (n/N = 15/716, 2.0%) and communicable diseases like pulmonary tuberculosis (n/N = 21/728, 2.9%), caused fewer deaths. Obstetric haemorrhage (n/N = 6/728, 0.8%) was the most common cause of maternal mortality, with pregnancy-induced hypertension and sepsis being other significant contributors. Table 2 outlines the ICD-10 coded CoD determinations made from VA data.
Table 2. WHO verbal autopsy causes of death according to ICD-10 Coding (n = 1500)
| WHO VA causes of death category | ICD Code | <5 y, n = 716 (%) | 5–18 y, n = 56 (%) | >18 y, n = 728 (%) |
|---|---|---|---|---|
| Non communicable diseases | ||||
| Acute cardiac disease | I24 | 5 (0.7) | 2 (3.6) | 182 (25.0) |
| Liver cirrhosis | K74 | 6 (0.8) | 3 (5.3) | 96 (13.1) |
| Stroke | I64 | 1 (0.1) | 1 (1.8) | 64 (8.8) |
| Renal failure | N19 | 5 (0.7) | - | 45 (6.1) |
| Other and unspecified disease | I99 | 21 (2.9) | 6 (10.7) | 20 (2.8) |
| Any other non-communicable disease | R99 | 8 (1.1) | 3 (5.3) | 18 (2.5) |
| Neoplasms | C06 | 1 (0.1) | 3 (5.3) | 25 (3.4) |
| Severe malnutrition | E46 | 4 (0.5) | - | 1 (0.1) |
| Cause of death unknown | R99 | 1 (0.1) | - | - |
| Diabetes mellitus | E14 | - | 1 (1.8) | 29 (0.3) |
| Respiratory neoplasms | C39 | - | 1 (1.8) | 10 (1.4) |
| COPD | J44 | - | - | 19 (2.6) |
| Oral neoplasm | C26 | - | - | 16 (2.2) |
| Acute abdomen | R10 | - | - | 8 (1.1) |
| Other and unspecified neoplasms | C80 | - | - | 5 (0.7) |
| Breast neoplasms | C50 | - | - | 4 (0.5) |
| Epilepsy | G40 | - | - | 3 (0.3) |
| Asthma | J45/J46 | - | - | 1 (0.1) |
| Female reproductive neoplasms | C57 | - | - | 1 (0.1) |
| Communicable diseases | ||||
| Diarrhoeal disease | A09 | 49 (6.8) | 4 (7.1) | 43 (6.0) |
| Pneumonia | J18 | 42 (5.8) | 4 (7.1) | 14 (1.9) |
| Sepsis | A41 | 34 (4.7) | 3 (5.3) | 14 (1.9) |
| Measles | B05 | 36 (5.0) | 3 (5.3) | - |
| Meningitis | G03 | 30 (4.1) | 1 (1.8) | 4 (0.5) |
| Other and unspecified infectious diseases | B99 | 5 (0.7) | 2 (3.6) | 7 (1.0) |
| Haemorrhagic fever (dengue) | A99 | 4 (0.5) | 6 (0.8) | |
| Pulmonary tuberculosis | A15 | - | 1 (1.8) | 21 (2.9) |
| Malaria | B54 | - | - | 4 (0.5) |
| External causes of death | ||||
| Road traffic crashes | V89 | 4 (0.5) | 5 (8.9) | 15 (2.1) |
| Accidental drowning and submersion | W74 | 5 (0.7) | 1 (1.8) | 5 (0.7) |
| Accidental fall | W19 | 2 (0.3) | 1 (1.8) | 4 (0.5) |
| Other and unspecified external causes | X59 | 3 (0.4) | 1 (1.8) | 2 (0.3) |
| Exposure to the force of nature | X39 | 1 (0.1) | - | 1 (0.1) |
| Accidental exposure to smoke, fire | X09 | - | 2 (3.6) | 1 (0.1) |
| Accidental poisoning and exposure to noxious substance | X49 | - | - | 9 (1.2) |
| Unknown | R00-R99 | 10 (1.3) | - | 14 (1.9) |
| Neonatal causes of death | ||||
| Neonatal sepsis | P63 | 91 (12.7) | - | - |
| Perinatal asphyxia | P21 | 84 (11.7) | - | - |
| Preterm birth complications | P07 | 52 (7.2) | - | - |
| Neonatal pneumonia | P23 | 27 (3.7) | - | - |
| Prematurity | P07.3 | 22 (3.0) | - | - |
| Congenital malformations | Q89 | 20 (2.7) | - | - |
| Cause not possible to determine from VA | P94 | 8 (1.1) | - | - |
| Other specific perinatal cause | P96 | 4 (0.5) | - | - |
| Term low birth weight | P07.1 | 3 (0.4) | - | - |
| Other obstetric complications | O75.4 | 3 (0.4) | - | - |
| Obstetric labour | O75 | 1 (0.1) | - | - |
| Still births | ||||
| Antepartum | O46.90 | 78 (10.8) | - | - |
| Intrapartum | O67.8 | 32 (4.4) | - | - |
| Other obstetric complications | O75.5 | 8 (1.1) | - | - |
| Antepartum haemorrhage | O4691 | 2 (0.3) | - | - |
| Cause not able to determine from VA | R69 | 2 (0.3) | - | - |
| Other medical conditions | R69 | 1 (0.1) | - | - |
| Cause of maternal mortality | ||||
| Obstetric haemorrhage (postpartum) | O72 | - | - | 6 (0.8) |
| Pregnancy-induced hypertension (pre-eclampsia) | O13 | - | - | 3 (0.4) |
| Other specific maternal causes | O05 | - | - | 1 (0.1) |
| Cause not possible to determine from VA | R69 | - | - | 1 (0.1) |
| Obstetric haemorrhage (antepartum) | O46 | - | - | 1 (0.1) |
| Other specific nonmaternal causes | O06 | - | - | 1 (0.1) |
| Pregnancy-induced hypertension (eclampsia) | O15 | - | - | 1 (0.1) |
| Pregnancy-related sepsis (antepartum) | O75.3 | - | - | 1 (0.1) |
| Unknown | R69 | - | - | - |
| Cause of death unknown | R99 | 4 (0.5) | - | 2 (0.3) |
COPD – chronic obstructive pulmonary disease, VA – verbal autopsy
Further, among children under five years (N = 524), neonatal sepsis (9.5% males, 7.8% females), perinatal asphyxia (8.6% males, 7.4% females), and antepartum complications (8.6% males, 6.1% females) were the leading causes of death. In the 5–18 years age group (n = 57), acute respiratory infections (10.5% for both genders), road traffic crashes (8.8% males, 1.8% females), and other unspecified diseases including congenital heart diseases (CHD) (5.3% males, 7.0% females) were prominent causes. For individuals older than 18 years (n = 471), acute cardiac disease (22.1% males, 1.5% females), liver cirrhosis (8.9% males, 11.3% females), and stroke (5.3% males, 8.3% females) accounted for the highest mortality (Figure 2).
Figure 2. Top 10 causes of death for the three categories. CHD – congenital heart diseases, COPD – chronic obstructive pulmonary disease.
DISCUSSION
The demographic insights from the VA study resonate deeply with the multifaceted nature of health care challenges in the country. Currently, the LMICs lack information regarding the specific CoD at the community level. Knowing the CoD at the population/community level is crucial since significant disparities exist between the CoD in the community and the medical setting. This study stands out as one of the few that comprehensively examined VA data for both adults and children, providing valuable insights into mortality that are essential for developing targeted public health interventions and improving health care outcomes across diverse demographic groups.
In this analysis, we have explored the distribution of death from a specific cause among adults and children and its distribution over time, sex, and age group. The present analysis showed nearly half of the death in adults were due to NCDs (41.2%), followed by deaths due to communicable diseases, neonatal causes etc. Our findings are partially consistent with analysis by the Global Burden of Disease 2019 which found ischemic heart disease, and stroke to be among the leading CoD in Pakistan [16]. In a country like Pakistan, with a population of over 200 million and approximately 190.27 million under the age of 40, there is a need to prioritise early detection, prevention, and management of NCDs, while maintaining robust surveillance and response systems for infectious diseases (Pakistan Population and Housing Census 2023).
Some LMIC studies like the Kersa health and demographic surveillance system (Kersa Health and Demographic Surveillance Systems) study of adult mortality in eastern Ethiopia, have reported lower rates of NCD associated mortality (26.4%) [18]. However, this difference may be due to differing geographical settings. Our study was conducted in an urban setting while the Kersa study population was predominantly composed of rural residents.
For a country such as Pakistan with a large percentage of the population aged under 40 years, efforts to reduce the burden of NCDs must focus on informing the youth and young adults about lifestyle choices, behaviours, and medical interventions that will reduce the risk of NCDs [16]. Although there is an aging population being increasingly affected by NCDs, investments in health systems and preventive medicine could prevent this from occurring in younger generations.
Furthermore, the leading CoD revealed in our study, such as acute cardiac disease, liver cirrhosis, diarrheal disease and pneumonia are consistent with Pakistan’s Global Burden of the Disease rates according to which ischemic heart disease and lower respiratory infections are leading CoD and contributing to a high overall mortality rate [16]. Our findings are also consistent with the Marsh DR et al. adult mortality study which examined 345 deaths between the ages of 15–59 years in a population of 45 389 in five slums in Karachi and where 55% of deaths were caused by cardiovascular diseases [19].
In our study, communicable diseases continued to remain a major cause of mortality, accounting for 23.2% of deaths. This is in line with a study conducted in Ethiopia where 30.3% of deaths were attributed to communicable diseases [20]. Other studies have also shown that the mortality burden from communicable diseases has decreased significantly over the past 30 years [21–23]. This might be due to the improvements in the health and socioeconomic status of the population over the years. In addition, a significant increase in the availability of primary health care services including vaccination could also be one factor in the observed finding.
In our study, neonatal deaths accounted for 21.2% of the total deaths, with neonatal sepsis being the predominating cause in this age group. Similarly in other LMICs, neonatal infection accounted for 17.6% of neonatal deaths [24]. The elevated rate of neonatal infections observed in this study could be attributed to inadequate health care-seeking behaviour among pregnant women. Many women seldom attend health care facilities for antenatal care, which is crucial for comprehensive maternal-foetal health management and timely detection and treatment of maternal infections. Similarly, during childbirth, many pregnant women still deliver at home and rarely seek postnatal care at health facilities [25]. This under-utilisation of health care facilities during the antenatal and peripartum periods results in inadequate access to specialised care necessary for newborns. A related factor could be the low level of female literacy in our populations. Consequently, mothers with limited education may struggle to recognise critical signs and symptoms of infections in newborns that require prompt medical attention. When symptoms are identified, many of these mothers tend to resort to local remedies for treatment [26].
Our study indicated that around one in 20 deaths (5.0%) were due to external causes with the majority being males i.e. 76.1%. This is in line with studies done in sub-Saharan Africa where the proportion of deaths from injury is higher among men as compared to women [27,28]. The variation in injury-related deaths between males and females can be attributed to two main factors: males often engage in activities that carry a higher risk of physical injury, and they are more frequently involved in interpersonal violence and deliberate self-harm [29–31]. In our study, the most economically active age-group (aged 18–50) are at the greatest risk of dying because of external CoD. This finding is similar to other studies in other developing countries [32]. The increased number of injury-related deaths among younger adults could suggest their heightened exposure to hazardous working conditions, such as in construction and other risky activities, compared to older adults.
Lastly, gender-specific patterns in CoD highlight nuanced health disparities. While acute cardiac disease remains a leading cause for both genders, gender-specific interventions may be needed to address certain health issues more effectively. The findings also emphasise the importance of integrated health care approaches that address both communicable and NCDs comprehensively. Strengthening primary health care systems, promoting health education and awareness, and addressing social determinants of health are crucial steps in improving health outcomes and reducing mortality rates in Pakistan.
Strengths and limitations
The VA study provides valuable insights into mortality patterns in Pakistan, but it comes with limitations that warrant consideration. One of the main concerns is the validity of the VA methodology, as it relies on retrospective information that may be subject to recall bias or misinterpretation of symptoms by family members or caregivers. This can affect the accuracy of CoD determinations, particularly for complex medical conditions or cases with incomplete information. This is particularly relevant in cases of neonatal and maternal deaths, where the symptom descriptions provided by family members may be limited, non-specific, or absent altogether due to cultural sensitivities, limited medical knowledge, or emotional distress.
Further, incomplete data and potential sampling bias limit generalisability, and the lack of a formal power calculation restricts the ability to detect fewer common CoD. Temporal factors and evolving health care access may also affect the findings’ relevance over time. Despite training, interviewer variability and subjective interpretation of responses may have introduced bias. Additionally, while physician review was used to assign CoD, inter-rater reliability metrics (e.g. Cohen’s kappa) were not calculated, limiting the assessment of diagnostic consistency. The exclusion of ethnicity from analytical models further constrained exploration of potential disparities.
Despite these limitations, the study also presents several strengths. One notable strength is its ability to capture data on CoD in a resource-limited setting where traditional death certification and autopsy data may be scarce. This enables the study to provide critical information about mortality patterns in areas where reliable data are often lacking. Furthermore, the use of standardised VA tools, when appropriately implemented, can improve the consistency of data collection and CoD attribution across diverse populations. The study’s findings have the potential to inform public health policies and interventions, guiding efforts to address the leading causes of mortality and improve health care delivery, particularly for vulnerable populations. Additionally, the collaboration with local health care systems and researchers enhances the study’s contextual relevance and ensures that it reflects the specific health care challenges faced in Pakistan.
CONCLUSIONS
The VA study offers valuable insights into mortality patterns and highlights the dual burden of NCDs and communicable diseases in Pakistan. These findings underscore the need for integrated strategies emphasising prevention, early detection, and stronger primary health care. While scaling up VA nationally could enhance mortality surveillance and inform health policy, its success hinges on overcoming key operational challenges such as workforce training, data standardisation, sustainable funding, and integration into existing health systems. Recognising and addressing these barriers is essential to fully realise the potential of VA as a tool for national health planning. Overall, VA remains a valuable method for informing evidence-based health care policies and interventions in Pakistan and can contribute meaningfully to improving health outcomes and well-being, particularly in under-resourced and underserved populations.
Acknowledgements
We would like to extend our heartfelt appreciation to the dedicated field workers who played an essential role in the success of this study. We are grateful to the families and communities who participated in this research, as their cooperation has been vital in providing insights into mortality patterns in Pakistan. Lastly, we would also like to express our sincere gratitude to Chrisitina Arif for her significant contributions to the study. Her support and involvement were instrumental in ensuring the smooth execution of various aspects of the research.