Tuberculosis and HIV/AIDS-attributed mortalities and associated sociodemographic factors in Papua New Guinea: evidence from the comprehensive health and epidemiological surveillance system.

OBJECTIVE: Tuberculosis (TB) and HIV/AIDS are public health concerns in Papua New Guinea (PNG). This study examines TB and HIV/AIDS mortalities and associated sociodemographic factors in PNG.
METHOD: As part of a longitudinal study, verbal autopsy (VA) interviews were conducted using the WHO 2016 VA Instrument to collect data of 926 deaths occurred in the communities within the catchment areas of the Comprehensive Health and Epidemiological Surveillance System from 2018 to 2020.InterVA-5 cause of deaths analytical tool was used to assign specific causes of death (COD). Multinomial logistic regression analyses were conducted to identify associated sociodemographic factors, estimate adjusted ORs (AOR), 95% CIs and p values. RESULT: TB and HIV/AIDS were the leading CODs from infectious diseases, attributed to 9% and 8% of the total deaths, respectively.Young adults (25-34 years) had the highest proportion of deaths from TB (20%) and the risk of dying from TB among this age group was five times more likely than those aged 75+ years (AOR: 5.5 (95% CI 1.4 to 21.7)). Urban populations were 46% less likely to die from this disease compared rural ones although the difference was not significant (AOR: 0.54 (95% CI 0.3 to 1.0)). People from middle household wealth quintile were three times more likely to die from TB than those in the richest quintile (AOR: 3.0 (95% CI 1.3 to 7.4)).Young adults also had the highest proportion of deaths to HIV/AIDS (18%) and were nearly seven times more likely to die from this disease compared with those aged 75+years (AOR: 6.7 (95% CI 1.7 to 25.4)). Males were 48% less likely to die from HIV/AIDS than females (AOR: 0.52 (95% CI 0.3 to 0.9)). The risk of dying from HIV/AIDS in urban population was 54% less likely than their rural counterparts (AOR: 0.46 (95% CI 0.2 to 0.9)).
CONCLUSION: TB and HIV/AIDS interventions are needed to target vulnerable populations to reduce premature mortality from these diseases in PNG. © Author(s) (or their employer(s)) 2022. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.

This study used tuberculosis (TB) and HIV/AIDS mortality data extracted from the Comprehensive Health and Epidemiological Surveillance System, providing data of more than 900 deaths recorded in the surveillance population, representing both urban–rural sectors of four geographical regions of Papua New Guinea.

Mortality data were collected in the period 2018–2020 via verbal autopsy (VA) interviews with close relatives of the deceased, who died in the communities, using the WHO 2016 VA interview instrument.

The InterVA-5 analytical tool was used to assign specific causes of death and categories in line with the International Classification of Diseases version 10.

TB and HIV/AIDS mortality data were linked with household socioeconomic data and household wealth index was constructed, allowing in-depth analyses of sociodemographic factors associated with mortalities from TB and HIV/AIDS.

Mortality data did not represent all the TB and HIV/AIDS deaths in the communities across the surveillance sites during the data collection period and might be biased due the recall process.

Background

The Sustainable Development Goal (SDG) 3.3 states: ‘By 2030, end the epidemics of AIDS, tuberculosis (TB), malaria, and neglected tropical diseases and combat hepatitis, waterborne diseases and other communicable diseases’1(2). TB and HIV/AIDS continue to be major global public health issues, having claimed almost 33 million lives. An estimated 38 million people are living with HIV and approximately 690 000 people died from HIV/AIDS, with an additional 1.7 million people newly infected in 2019.2 In the Western Pacific Region, TB remains a major public health concern, accounting for nearly 20% of the global burden with an estimated 1.8 million new cases reported in 2019.3

Papua New Guinea (PNG)4 5 is the largest nation in the South Pacific region with a total population of approximately 8 million, and annual population growth rate of 2.8% in the decade 2000 s-2010s. Life expectancy at birth was at 63 years in 2010 and 40% of the PNG population are under 15 years of age.4 6 Rural populations account for more than 85% of the entire population and are widely scattered across the four geographical regions: Highlands, Southern, Momase and Islands. PNG is classified as a lower-middle-income country and a signatory to the SDGs.5 PNG has recently undergone an epidemiological transition with premature mortality continuing to decline from 2010 to 2020.7 Infectious diseases are still the dominant cause of morbidity, accounting for almost half the total burden of diseases and illnesses at the primary health level.8 9

TB and HIV/AIDS were reported as major public health concerns in PNG in the 2010s.10 11 PNG had the highest HIV/AIDS incidence and prevalence in the Pacific region accounting for 95% of the reported HIV/AIDS cases in the region.12 This disease has had severe impacts on the health sector and it was estimated that people living with AIDS occupied 70% of hospital beds in 2015.13 With TB prevalence infection rate of approximately 333 cases per 100 000 population, PNG was classified among 14 countries with the highest burden of TB in the world in 2016. TB prevalence was particularly high among vulnerable populations such as female sex workers and men who have sex with men in Port Moresby, about 1200 and 1000 per 100 000 in 2017, respectively.14 It was estimated in 2018 that 37 000 people contracted TB and around 4500 of these people died from this disease.3 5 15 As of 2020, approximately 45 000 people living with HIV/AIDS were reported in PNG,2 with this figure likely to be underreported.16

In PNG, HIV/AIDS and TB interventions and services are integrated into public health services at the local level and under the administration of Provincial Health Authorities (PHA), which were established across provinces under the PHA Act 2007. The issue of this Act was part of the government decentralisation to increase accountability for the provincial and local level governments to improve the standard of public health practices and the delivery of public health services, including preventive and curative services to communities through provincial health partnerships.15 The national HIV/AIDS and TB programmes provide antiretrovirus therapy (ART) and anti-TB drugs among other services under the 2016–2020 National Health Plan.17

Major challenges in the delivery of TB and HIV/AIDS prevention and control programmes include underdetection of new cases, poor treatment outcomes and the high numbers of TB and HIV/AIDS patients who were lost to follow-up.18 TB and HIV/AIDS patients can seek healthcare services at health facilities in urban areas. These patients can be admitted to a tertiary hospital for treatment, but they are often discharged from the hospitals in the late stage of their diseases and die at home in their own villages. Hence the records of TB and HIV/AIDS deaths that occur in the communities are more likely completed than those recorded in health facilities, particularly in the rural areas, where access to TB and HIV/AIDS services are limited.

Social determinants of TB and HIV/AIDS mortality are poorly understood despite of the heavy burden of these diseases on health systems in PNG. Study of social determinants of mortality from TB and HIV/AIDS is important for public health policy and interventions.19 Understanding of social determinants of TB and HIV/AIDS mortalities provides insight into the performance of health systems and healthcare interventions in reducing mortality from these diseases, contributing to achievement of SDGs. In African countries, higher premature mortality from HIV/AIDS was reported among females compared with males, people living in rural areas and those from lower household socioeconomic status (SES).20 Sociodemographic factors such as age, sex, education, marital and employment status, housing condition and household SES were also associated with HIV/AIDS mortality.21–24 The impact of TB and HIV/AIDS on mortality of a population could be likely detected and identified at the individual and household levels. Household socioeconomic demographic factors should be considered when examining social determinants of mortality. However, few studies on household and individual sociodemographic factors associated with mortalities attributed to TB and HIV have been conducted in countries in the Western Pacific region. No known study has examined sociodemographic factors related to mortality from TB and HIV/AIDS in PNG.

This study examined the proportion of mortality from TB and HIV/AIDS among people in PNG and explored the possible associations of key sociodemographic factors with these mortalities. The study aimed to address the following research questions:

What are the distributions of mortality from TB and HIV/AIDS by age, sex and household SES of the deceased and by urban–rural sectors and provinces?

What are the sociodemographic factors associated with TB attributed mortality?

What are the sociodemographic factors associated with mortality attributed to HIV/AIDS?

Materials and methods

Data source

Mortality surveillance data were extracted from the Comprehensive Health and Epidemiological Surveillance System (CHESS), operated since 2018 by Papua New Guinea Institute of Medical Research (PNGIMR). CHESS was based on the integrated Health and Demographic Surveillance System (iHDSS), which was established in PNG in the period 2010–2017. CHESS was designed as a population-based longitudinal follow-up cohort system. The overall purpose of CHESS was to provide a reliable and up-to-date data series for monitoring the implementation of socioeconomic development programmes and healthcare interventions at the subnational level in PNG. CHESS catchment areas include eight surveillance sites located in six provinces: Eastern Highlands Province (EHP), East New Britain (ENB), East Sepik Province (ESP), Central, Madang and Port Moresby (POM—the National Capital District). By the end 2022, CHESS will cover a population size of approximately 80 000, equivalent to 1% of the total population of PNG. The system provides population data from rural and urban sectors, with approximately 75% of rural and 25% of urban populations, comparable with the national rural–urban population distribution for the period 2018–2022.4 The designs and methods of iHDSS and CHESS have been previously published.25 26 The distance between urban-rural sites in EHP and ENB is about 50 km. This provides a balance between facilitating access and transportation and ensuring differences in socioeconomic development can be observed and captured in the data. Sociodemographic characteristics of the surveillance population by sites are presented in table 1.

Table 1

Socioeconomic characteristics of the surveillance sites, PNGIMR’s CHESS, 2018–2020

Province	Port moresby	Central	Eastern highlands	Eastern highlands	Madang	East sepik	East New Britain (ENB)	ENB
Surveillance site	Hohola	Hiri	Goroka	Asaro	Newtown	Maprik	Kokopo	Baining
Sector	Urban	Rural	Urban	Rural	Urban	Rural	Urban	Rural
Region	Southern	Southern	Highlands	Highlands	Momase	Momase	Islands	Islands
Location	National Capital District	45 km west of Port Moresby	Township of EHP	50 km northeast of Goroka	Township of Madang	30 km from township	Township of ENB	40 km from the town
Main industry	Shipping, transportation	Fishery, hunting	Coffee, agriculture	Coffee, agriculture	Fishery, services	Vanilla, cocoa	Fishery, tourism	Fishery, tourism
Accessibility	Road and airline	Road	Road and airline	Road and airline	Road and airline	Road and airline	Sea and airline	Sea and airline
Year of site established	2017	2011	2016	2004	2018	2019	2018	2018
Population	5000	15 000	5000	15 000	5000	5000	5000	6000
Household	1000	3000	1000	3000	1000	3000	1000	3000
Health facility	St. Theresa clinic	Porebada, Papa and Lealea clinics	Provincial Hospital Kwongi, Lopi and Goroka clinics:	Asaro Health Centre	Jomba Clinic	Ilahita clinic,District Hospital	Batuwin Clinic	Vanapalading Aid Post
Laboratory services	POM Lab	N/A	Goroka Lab	N/A	Madang Lab	N/A	N/A	N/A

CHESS, Comprehensive Health and Epidemiological Surveillance System; NA, non-available; PNGIMR, Papua New Guinea Institute of Medical Research; POM, Port Moresby.

Data collection

Mortality surveillance data were collected from the population living in the CHESS sites in the period 2018–2020, using the WHO 2016 verbal autopsy (VA) interview instrument. This tool is based on the consolidation of various existing VA tools and programmed for conducting VA interviews using portable electronic devices.27 The WHO 2016 tool does not require interviewers to have a health and medical background to conduct VA interviews.28 The WHO 2016 VA instrument was adapted in 2017 for optimal use in the local context and integrated into CHESS surveillance activities in 2018.29 An additional data module on identification information of the deceased, including household global positioning system (GPS) data and individual ID code was included in the VA instrument for this study, allows linkage between mortality and household socioeconomic demographic data.

The field work and data collection were integrated into the ongoing routine surveillance activities of the CHESS in PNG. The mortality data were collected from the surveillance population, who live in eight surveillance sites established across six provinces: Central, Port Moresby (POM—the National Capital District), Eastern Highlands, Madang, East Sepik and ENB. Data used in this study focused on deaths from the communities and no death records from health facilities were included.

Deaths in the communities were identified by data reporters, who are local people living in their villages, recruited to work for CHESS, and based in their villages. They collected information on birth, death and migration through regular visits to households for collecting information on demographic changes. Given the social network, data reporters were easily aware of deaths that occurred in their villages and had access to the households at a convenient time to collect further information about the deceased, including the date of death. Data reporters prearranged VA interviews at a time and location that convenient for both interviewer and interviewee to attend.

The mortality data and information were collected from March 2018 to September 2020. VA interviews were conducted by national scientific officers of the CHESS’s demographic team in Tok-Pisin, the most common local language in PNG and Motu language was used in Central Province. VA interviews were usually scheduled in the 2 weeks after mourning period.30 However, the organisation of VA interviews could take several weeks due to logistical arrangements and the availability of interviewees and transportation means. Some VA interviews required more than one visit to complete. The completion of VA interviews was also prolonged because of lockdowns for several periods and CHESS staff members being infected with COVID-19 during the COVID-19 outbreaks in 2020.

Household relatives, who participated in VA interviews, were often household heads for adults who were deceased and parents/caregivers for child who were deceased. These participants should have spent a considerable period of time to directly take care of the deceased prior to the deaths, who were able to recognise and remember important clinical signs the deceased demonstrated in their last stage prior to deaths, who were capacitate, willing to cooperate in VA interview and respond to the interview questions.

Data linkage

An additional data module on the deceased identification information was included in the questionnaire, including household location (GPS data) and individual identification information. This information allows identifying the deceased in the communities and linkages their mortality data and other existing data components available from the CHESS database, including morbidity data and household socioeconomic data.

The household and individual ID coding systems used in CHESS were created in 2014–2015, aligned with the national coding system, published by the National Statistics Office.4 Household codes consist of 17 digits representing for province/city (2 digits), district/town (2 digits), local level government (2 digits), commune/ward (2 digits), village/street (3 digits), dwelling/compound (3 digits) and household number (3 digits). The three last code identifiers were established by the CHESS. Individual ID codes were constructed based on the household ID code by adding two digits (household individual line number) to the end of their respective household ID codes. Individual ID codes are updated on regular basis, using the household demographic change data on birth, death and migration in and out of the households. In addition, households can be also identified, using household GPS data on latitude (Degree South with eight digits), longitude (Degree East with eight digits) and elevation (metre with six digits).

These coding systems are applied consistently across the surveillance sites and studies conducted using the CHESS research platform that allows identifying households and individuals participating in different studies. Mortality data and household socioeconomic demographic data were linked together by using the household and individual ID codes of the deceased. The linked mortality-household SES data set was for use in the analyses of sociodemographic factors in this study.

The CHESS database is updated with household SES data on every 2 years. The SES data used in this study were also collected from January to June 2018 by village-based data reporters under supervision of demographic team leaders and site managers. Household interviews were conducted with household members, most often with household heads, using the household SES questionnaire, which was composed of nine data modules: (1) Household identification information, including GPS; (2) List of household members and their relationship to household head; (3) Education level of household members aged 5 or above; (4) Employment status of household members of working age 15–64; (5) Access, availability and utilisation of bed-nets; (6) Water and sanitation; (7) Hand washing; (8) Housing characteristics and household assets and (9) Access and utilisation of health services. These data were used to construct the household wealth index.

Data analyses

The InterVA-5 COD analytical tool was used to analyse causes of death (CODs) using VA interview data. This computer-based programme can assign 64 specific CODs and categories in line with the International Classification of Diseases version 10.31 Among the 1021 deaths identified in the communities, consents were obtained for conducing 1003 VA interviews, resulted in a participation rate of 98%. InterVA-5 COD analytical tool successfully assigned specific CODs for 926 VA interviews. InterVA-5 programme can assign more than one specific COD for a death with respective likelihoods. However, in this study, only the first ascribed CODs with the highest likelihoods were analysed. For instance of deaths when TB was assigned as the first COD and HIV was the second COD, only TB was included in the TB-attributed mortality analysis. Similarly, in the case when HIV was assigned as the first COD and TB was the second COD, then only HIV attributed deaths were included in the HIV mortality analysis

To analyse mortalities from TB and from HIV/AIDS by selected sociodemographic factors, VA data were linked with the household socioeconomic (SES) data from the corresponding period of time using the unique household and individual identification codes. Specifically, the 2018–2020 VA data were linked with the 2018 household SES data. Mortality data from 665 deaths were successfully linked with household SES data and included in the analyses. No household SES data for ESP for 2018 were available as the site was not established until early 2019.

A new variable on household wealth index was constructed for each deceased using the principal component analysis (PCA) method. The application of PCA in the PNGIMR’s CHESS has been previously published.32 Household SES and demographic variables were included in PCA models. Significant variables remained in the PCA model including housing characteristics, water and sanitation, and household assets. Non significant variables were excluded from the models including education, employment, and occupation of the deceased. Household wealth indices were then divided into quintiles and categorised as poorest, poor, middle, richer and richest.

Two binary variables were created: (1) TB attributed death (‘Yes’ was for deaths from TB and ‘No’ was deaths from any other COD (infectious or non-communicable diseases) and (2) HIV/AIDS attributed death (‘Yes’ was deaths from HIV/AIDS and ‘No’ was deaths from any other COD (infectious or non-communicable diseases). These variables were included in logistic regression analyses as dependent variables, and sociodemographic factors were independent variables.33

Unadjusted and adjusted ORs (AORs) of mortalities from TB and HIV were first produced by using the binary logistic regression analysis. All significant variables identified in these analyses were then included in multinomial logistic regression (MLR) to predict the increased risk of mortalities from TB and from HIV/AIDS across subpopulations. The significant variables remained in the MLR models including age at death, sex of the deceased, household wealth quintile and urban–rural sector, except for the ‘province’ variable, which was excluded in the model because of confounding with the urban–rural sector variable (The surveillance site in Port Moresby is located urban area while the site in Central Province in rural area). Main effect was selected to produce estimates of ORs for the risks of dying from TB and HIV/AIDS. Statistical likelihood tests were used to provide 95% CIs of the estimated ORs. A p<0.05 was considered as significant. All analyses were performed using the Statistical Package for Social Sciences (SPSS V.20).

Patient and public involvement

No patient involved.

Results

TB and HIV/AIDS as top leading infectious disease COD

Figure 1 shows mortality from infectious diseases among the surveillance population. A total of 317 deaths were attributed to infectious diseases, in which pulmonary TB was the leading COD and responsible for 84 deaths, followed by acute respiratory tract infections (81 deaths) and HIV/AIDS (75 deaths). Among the 84 deaths attributed to TB, 3 deaths assigned HIV/AIDS as the second COD, accounting for 3.6% of TB deaths. Similarly, among the 75 deaths from HIV/AIDS, 6 deaths assigned TB as the second COD, accounting for 0.8% of HIV/AIDS deaths. Hence the comorbidities of TB and HIV/AIDS could be about 6% of all HIV/AIDS and TB deaths.

Figure 1

Leading causes of deaths from infectious diseases (317 deaths) in the communities in PNG, PNGIMR’s CHESS, 2018–2020. CHESS, Comprehensive Health and Epidemiological Surveillance System; PNG, Papua New Guinea; PNGIMR, Papua New Guinea Institute of Medical Research; RTI, Respiratory Tract Infections.

Table 2 shows the distribution of TB and HIV/AIDS attributed mortalities by sociodemographic characteristics of the deceased. The mean age of death from TB was 46.3 (±17.58), which was slightly higher than for HIV/AIDS, 44.21 (±19.36), but lower than for other CODs, 49.24 (±24.32) (Eta squared p=0.04). TB accounted for 9% of total deaths in the surveillance population. TB claimed the highest number of deaths in those aged 25–34, comprising 20% of all CODs recorded in this age group. This was followed by the 35–44 years age group with 18% of deaths. The highest number of TB deaths was observed among people from households in the middle quintile. A slightly higher proportion of deaths occurred in rural areas (10%) compared with urban areas (7%). Central province had the highest numbers of TB deaths, with 41 deaths which accounted for 14% of the death records in this province.

Table 2

Distribution of deaths from pulmonary TB, HIV/AIDS and other causes of death (number and percents) by age group, sex, urban–rural sector, province, and household wealth quintile of the deceased, PNGIMR’s CHESS, 2018–2020

		TB	HIV/AIDS	Other CODs	All CODs
Mean age at death (year, SD)	P value: 0.04	46.32 (17.58)	44.21 (19.36)	49.24 (24.32)	48.56 (23.44)
Age group	0–4	0 (0.0%)	3 (4.5%)	64 (95.5%)	67 (100.0%)
	5–14	1 (3.8%)	0 (0.0%)	25 (96.2%)	26 (100.0%)
	15–24	7 (11.3%)	6 (9.7%)	49 (79.0%)	62 (100.0%)
	25–34	20 (20.2%)	18 (18.2%)	61 (61.6%)	99 (100.0%)
	35–44	17 (18.1%)	11 (11.7%)	66 (70.2%)	94 (100.0%)
	45–54	9 (6.0%)	14 (9.4%)	126 (84.6%)	149 (100.0%)
	55–64	15 (8.8%)	14 (8.2%)	141 (82.9%)	170 (100.0%)
	65–74	12 (8.4%)	4 (2.8%)	127 (88.8%)	143 (100.0%)
	75+	3 (2.7%)	5 (4.4%)	105 (92.9%)	113 (100.0%)
Total		84 (9.1%)	75 (8.1%)	764 (82.8%)	923 (100.0%)
Sex	Male	43 (8.4%)	28 (5.4%)	443 (86.2%)	514 (100.0%)
	Female	41 (10.0%)	47 (11.4%)	324 (78.6%)	412 (100.0%)
Total		84 (9.1%)	75 (8.1%)	767 (82.8%)	926 (100.0%)
Sector	Urban	16 (7.0%)	13 (5.7%)	199 (87.3%)	228 (100.0%)
	Rural	67 (9.9%)	60 (8.9%)	550 (81.2%)	677 (100.0%)
Total		83 (9.2%)	73 (8.1%)	749 (82.8%)	905 (100.0%)
Household wealth quintile	Poorest	9 (6.5%)	14 (10.1%)	115 (83.3%)	138 (100.0%)
	Poor	7 (5.1%)	15 (10.9%)	116 (84.1%)	138 (100.0%)
	Middle	21 (15.2%)	14 (10.1%)	103 (74.6%)	138 (100.0%)
	Rich	16 (11.6%)	10 (7.2%)	112 (81.2%)	138 (100.0%)
	Richest	8 (5.8%)	8 (5.8%)	121 (88.3%)	137 (100.0%)
Total		61 (8.9%)	61 (8.9%)	567 (82.3%)	689 (100.0%)
Province	Port Moresby	1 (3.3%)	2 (6.7%)	27 (90.0%)	30 (100.0%)
	Central	41 (14.2%)	13 (4.5%)	234 (81.3%)	288 (100.0%)
	Eastern Highlands	25 (8.3%)	36 (12.0%)	239 (79.7%)	300 (100.0%)
	Madang	8 (10.5%)	8 (10.5%)	60 (78.9%)	76 (100.0%)
	East Sepik	2 (1.7%)	8 (6.9%)	106 (91.4%)	116 (100.0%)
	East New Britain	7 (6.0%)	8 (6.9%)	101 (87.1%)	116 (100.0%)
Total		84 (9.1%)	75 (8.1%)	767 (82.8%)	926 (100.0%)

CHESS, Comprehensive Health and Epidemiological Surveillance System; CODs, causes of death; PNGIMR, Papua New Guinea Institute of Medical Research; TB, tuberculosis.

HIV/AIDS attributed to 8% of the total deaths recorded in the surveillance population. The proportion of deaths attributed to HIV/AIDS was more than twice for females than males, 11% and 5%, respectively. The population aged 25–34 years had the highest number of deaths from HIV/AIDS, accounting for 18% of deaths in this age group. Three HIV/AIDS deaths were identified among children aged 0–4, accounting for 5% of deaths among children in this age group. There were 60 HIV/AIDS deaths identified in rural population, accounting for 9% of deaths in rural areas, compared with 6% reported in urban areas. EHP had the highest number of deaths from HIV/AIDS, with 36 deaths that accounted for 12% of deaths recorded in this province. By contrast, Central province recorded the lowest proportion of deaths from HIV/AIDS (4.5%).

Socioeconomic demographic factors of mortalities from TB and HIV/AIDS

Table 3 shows the adjusted ORs of mortality from TB by sociodemographic characteristics of the deceased (only adjusted ORs are presented because the unadjusted and adjusted ORs were similar). Those aged 25–34 were over five times more likely to die from TB than those aged 75+years (AOR: 5.48 (95% CI 1.38 to 21.75)). Urban populations were 46% less likely to die from TB than those in rural areas (AOR: 0.54 95% CI 0.28 to 1.0)) although the difference was not significant (p=0.05). People from the middle household wealth quintile were three times more likely to die from TB than those from the richest quintile (AOR: 3.06 (95% CI 1.27 to 7.37)). The difference in TB mortality between males and females was not significant (p=0.47).

Table 3

Distribution of deaths and adjusted ORs of mortality from pulmonary tuberculosis vs all other causes of death, by sociodemographic characteristics of the deceased, multinomial logistic regression analysis, PNGIMR’s CHESS, 2018–2020

Sociodemographic characteristics	Category	N	%	Adjusted OR	Lower bound	Upper bound	P value
Age group (in year)	0–4	39	5.9%	NA	NA	NA	NA
	5–14	20	3.0%	NA	NA	NA	NA
	15–24	45	6.8%	2.982	0.621	14.317	0.172
	25–34	64	9.6%	5.482	1.382	21.751	0.016
	35–44	62	9.3%	6.428	1.675	24.664	0.007
	45–54	115	17.3%	2.290	0.582	9.009	0.236
	55–64	131	19.7%	3.118	0.856	11.360	0.085
	65–74	102	15.3%	3.438	0.913	12.955	0.068
	75–102	87	13.1%	Ref.
Sex	Male	387	58.2%	1.227	0.696	2.164	0.479
	Female	278	41.8%	Ref.
Sector	Urban	226	34.0%	0.540	0.288	1.000	0.050
	Rural	439	66.0%	Ref.
Household wealth	Poorest	135	20.3%	0.997	0.367	2.708	0.995
	Poor	135	20.3%	0.817	0.283	2.359	0.709
	Middle	130	19.5%	3.067	1.275	7.374	0.012
	Rich	130	19.5%	2.005	0.802	5.010	0.137
	Richest	135	20.3%	Ref.
Valid total		665	100.0%

Dependent variable was deaths from tuberculosis. Reference category was deaths from other CODs. Sociodemographic factors remained in MLR models included age, sex, urban–rural sector and household wealth.

CHESS, Comprehensive Health and Epidemiological Surveillance System; CODs, causes of death; MLR, multinomial logistic regression; NA, non-available; PNGIMR, Papua New Guinea Institute of Medical Research.

Table 4 shows the AORs of mortality from HIV/AIDS by sociodemographic characteristics of the deceased (only AORs are presented because the unadjusted and adjusted ORs were similar). Similar to TB, those aged 25–34 were nearly seven times more likely to die from HIV/AIDS than those aged 75+ years (AOR: 6.68 (95% CI 1.75 to 25.43)). Males were about 50% less likely to die from HIV/AIDS than their female counterparts (AOR: 0.52 (95% CI 0.29 to 0.9)). Urban population were about 55% less likely to die from HIV/AIDS than rural populations (AOR: 0.46 (95% CI 0.24 to 0.89)). The differences in HIV/AIDS mortality were not significant among household wealth quintiles (p>0.05).

Table 4

Distribution of deaths and adjusted ORs of mortality from HIV/AIDS versus all other causes of death, by sociodemographic characteristics of the deceased, multinomial logistic regression analysis, PNGIMR’s CHESS, 2018–2020

Sociodemographic characteristics	Category	N	%	Adjusted OR	Lower bound	Upper bound	P value
Age group	0–4	39	5.9%	1.511	0.240	9.531	0.660
	5–14	20	3.0%	NA	NA	NA	NA
	15–24	45	6.8%	3.461	0.771	15.548	0.105
	25–34	64	9.6%	6.687	1.758	25.438	0.005
	35–44	62	9.3%	4.872	1.240	19.150	0.023
	45–54	115	17.3%	2.868	0.765	10.758	0.118
	55–64	131	19.7%	2.876	0.783	10.565	0.112
	65–74	102	15.3%	1.108	0.238	5.150	0.896
	75–102	87	13.1%	Ref.
Sex	Male	387	58.2%	0.517	0.294	0.908	0.022
	Female	278	41.8%	Ref.
Sector	Urban	226	34.0%	0.464	0.240	0.899	0.023
	Rural	439	66.0%	Ref.
Household wealth	Poorest	135	20.3%	2.084	0.797	5.452	0.134
	Poor	135	20.3%	2.342	0.901	6.090	0.081
	Middle	130	19.5%	1.999	0.751	5.319	0.165
	Rich	130	19.5%	1.550	0.557	4.311	0.401
	Richest	135	20.3%	Ref.
Valid		665	100.0%

Note: Dependent variable was deaths from HIV/AIDS. Reference category was deaths from other CODs. Sociodemographic factors remained in MLR models included age, sex, urban-rural sector and household wealth.

CHESS, Comprehensive Health and Epidemiological Surveillance System; PNGIMR, Papua New Guinea Institute of Medical Research.

Discussion

Using the linked dataset between mortality and household SES data from the CHESS database, we have conducted analyses to identify that TB and HIV/AIDS were the leading CODs from infectious diseases in the population in PNG. Analysis of key sociodemographic factors of mortalities from these diseases, we found that age, sex, urban–rural residence and household wealth had significant associations, but the effects of these factors on TB and HIV/AIDS mortalities were varied. People in the age groups 25–34 and 35–44 years were at the highest risk to die from both TB (p values of 0.016 and 0.007, respectively) and HIV/AIDS (p values of 0.005 and 0.023, respectively) (see tables 3 and 4). Females and those living in rural areas were more at risk to die from HIV/AIDS than males and urban population (p =value of 0.023, see table 4). People from households in middle wealth quintile were more likely to die from TB than those from the highest quintile (p0.012, see table 3), but there was no significant association between HIV/AIDS mortality and household SES (p value above 0.05 across the wealth quintiles, see table 4).

We found that PNG people were dying from TB and HIV/AIDS at a very young age, with the highest premature mortality among the young adults. Specifically, TB and HIV/AIDS were responsible for nearly two in every five deaths (or 38% of all deaths) in the population aged 25–34 years. From our observations, young people living in the surveillance sites are highly mobile. Many regularly move from one place to another for education, employment, social and family purposes. They are also sexually active and more likely involved in unprotected sex.34 Lacking access to preventive measure such as condom, young people are more likely to become infected with HIV/AIDS among other sexually transmitted infections. Young people are the main labour force and the most productive in the national and household economies. The loss of young people to TB and HIV/AIDS present a significant economic cost to their families, the communities and society as a whole.34 It was estimated if there were 300 000 adult deaths to HIV/AIDS; the workforce would be reduced by 12.5%; and the annual GDP growth rate would decline by 1.3% by 2025.13

Gender inequality was a key factor associated with the increased risk of dying from HIV/AIDS among the female population. In PNG, women are culturally considered as having a lower social status than men. Young women are more likely to engage in high risk sexual activities to satisfy or meet their needs and wants than men. Given the low prevalence of safe sex practices, these women are more likely to be exposed to HIV/AIDS infection.35 In most cases, women are more likely to die younger than men if they are infected with HIV/AIDS. HIV/AIDS also attributed to 5% of deaths among children under 5 years of age. HIV/AIDS prevention of mother-to-child transmission (PMTCT) programmes have been reportedly integrated into antenatal care services in PNG, but access to these services is limited. Because of social stigma and discrimination, many HIV/AIDS pregnant women did not give correct personal information,35 further hindering the utilisation of antenatal care services. The loss of mothers is likely to leave a large socioeconomic burden to their families, and children are particularly impacted from the loss.18 Infant children died from HIV/AIDS are evidence of the failure of delivery of the PMTCT programme in PNG. Reducing social stigma and discrimination against people living with HIV/AIDS, particularly women is crucial to improving the delivery of HIV/AIDS services. Increasing men’s roles in the national response to HIV/AIDS prevention is also important for a successful implementation of HIV/AIDS programme in PNG.

Our study has shown that people who live in the rural areas were twice as likely to die from TB and HIV/AIDS as those in urban areas. Inadequate health promotion and education could be the cause of low public awareness about the diseases and poor knowledge and practices towards prevention of TB and HIV/AIDS among rural populations. Limited access to basic healthcare services is often cited as the main reason for the high mortality from TB and HIV/AIDS in PNG.36 37 The unavailability of skilled health workers, lack of essential drugs and consumables are often reported at primary health facilities in rural areas.10 Access to ART has been a key to reducing the risk of dying from HIV/AIDS, but this service is available only at a small number of tertiary health facilities such as hospitals in Port Moresby. Essential laboratory services such as GeneXpert and TB culture for monitoring multidrug-resistant-TB (MDR-TB), and HIV testing are very limited, even unavailable on regular basis at the EHP Hospital. No community-based modality is available for effective detection and management of TB and HIV/AIDS cases.38 Control of the spread of TB and HIV/AIDS infections in the communities has been ineffective due to lost to follow-up with the patients.14 35

The high mobility among young populations between urban and rural areas and particularly the increased number of young migrants moving from rural to urban areas for employment would further complicate the spread of HIV and TB, elevating the higher risks of dying from these diseases, particularly in rural areas, where the access to HIV/AIDS services were even more limited. Health education on preventive measures of HIV/AIDS, including safe sex practice and equitable access to youth-friendly HIV and TB services are needed to reduce the premature mortalities among young people

TB-HIV coinfections have increased and are one of the key challenges to the effective implementation of TB and HIV/AIDS programmes in PNG. Our study estimated that the prevalence of TB and HIV/AIDS coinfections could be about 6% of HIV/AIDS and TB patients. This finding is consistent with our morbidity surveillance data.37 The recent emergence of HIV/AIDS coinfections with TB in Western and EHP has raised a public health threat of MDR-TB.17 38 The resurgence of tropical neglected infectious diseases such as typhoid,39 leprosy,38 recent outbreaks of childhood communicable diseases including polio and measles,40 and the current spread of COVID-19 infection in the communities have imposed threats to the overwhelmed healthcare systems.

Using the update-to-date data extracted from the CHESS database, we were able to link the mortality data and the household SES data to enhance the scope of mortality analyses, especially in analysing the sociodemographic factors associated with the increased risk of dying from TB and HIV among the population (The list of significant variables retained in the CPA model component 1 for constructing household wealth index is shown in online supplemental table 1). The use of WHO 2016 VA tool to collect mortality data from the communities are more likely to reflect a real and complete picture of mortality in the population than health facility death records. The InterVA-5 analytical programme was used for the first time in this study in PNG to ascertain CODs from TB and HIV in the population that can be scaled up in PNG and replicated in similar settings

The mortality data were collected via VA interviews with close relatives of the deceased. Although the WHO 2016 VA instrument had been pretested with the local people prior to the data collection, the provided information about the deaths may be incomplete and biased due the recall process. InterVA-5 is a standard tool, but the ascribed CODs could be also biases due to the death selection process. The mortality data were collected from the population living within the CHESS catchment areas, but the data included only deaths identified by the village-based data reporters. It is challenging to ensure all deaths occurred in the communities were included in the data. Hence, the data used were not representative for all deaths in the surveillance sites across provinces.

Given the high level of social stigma around TB and discrimination against people living with HIV/AIDS in PNG, deaths from these diseases might have been under-reported in this study. The relatively small numbers of deaths, 84 from TB and 75 from HIV compared with all other CODs might have hindered the identification of sociodemographic factors of TB and HIV attributed mortalities. The limited numbers of observations could have also led to important associations being non-significant when TB and HIV deaths were compared with deaths from all other causes (unadjusted and AORs of TB and HIV/AIDS combined mortalities in binary logistic regression models are presented in online supplemental table 2), which are different from those presented in tables 3 and 4, where AORs of TB mortality and HIV mortality were estimated separately). The small sample size may have limited ability to draw concrete conclusions based on AORs that were estimated with large CIs covering node value of 1 and non-significant p values of 0.05 or above. This limits the interpretation of risk factors associated TB and HIV mortalities in this study. An assessment of the accuracy and reliability of specific CODs assigned by InterVA-5 is beyond the scope of this study and will be addressed in a separate study.

Conclusions

Over the past 40 years, HIV/AIDS infections have transitioned to a manageable chronic infection in many LMICs and TB infection has been contained in many parts of the world. The current high mortality from TB and HIV/AIDS in PNG appears in contrast to the global trend of declining infections and deaths from these diseases. TB and HIV/AIDS have recently emerged in PNG and becoming leading COD in the population. The high premature mortalities from TB and HIV/AIDS among young people, together with the increased TB-HIV coinfections have raised public concerns about the TB and HIV/AIDS programmes in PNG, threatening the sustainable development of the country.8 41 42

Urgent actions are needed from the PNG Government and health sector to review the current strategies and plans for further improvement in the effectiveness of TB and HIV/AIDS programmes as well as the delivery of healthcare services to the population in PNG in the next decade. More interventions are needed with focus on high-risk and vulnerable populations, particularly those who are young and females in rural areas and from provinces with low socioeconomic development status. The identified sociodemographic factors associated with premature mortality attributed to TB and HIV/AIDS need to be tackled. Further studies on the trend of mortality transition and the change in CODs across different social classes are needed to better inform policy and intervention.