Health system inequities in Lao People's Democratic Republic: Evidence from a nationally representative phone survey

Despite substantial economic growth in Lao People's Democratic Republic (PDR) over the past 20 years, high levels of income inequality and poverty persist and have likely been exacerbated by the COVID‐19 pandemic. In this article, we use novel survey data to assess the extent to which socioeconomic status is associated with access to quality care in Lao PDR.


INTRODUCTION
Although the Lao People's Democratic Republic (Lao PDR) has achieved substantial economic gains and improvements in population level health outcomes, significant gaps in equity persist.During the period of 2000-2020, the Lao PDR economy grew at a rather remarkable average annual rate of 7.9% [1].The growth, however, has not benefitted all population groups equally with persistently high poverty rates among ethnic minority groups and rising income inequality [2].These socioeconomic differences are also visible in terms of health outcomes with poor and ethnic minorities having shorter life expectancies and carrying a disproportionate burden of maternal and child mortalities [3,4].Under-five mortality rates were nearly double for Mon-Khmer ethnic group (63 deaths per 1000 live births) and Chinese-Tibetan ethnic group (72 deaths per 100 live births) compared to the Lao-Tai majority (35 deaths per 1000 live births) [5].Similarly, households in the poorest quintile had more than 2.5 times the under-5 mortality rates (63 deaths per 1000 live births) of households in the richest quintile (23 deaths per 1000 live births) [5].
The Government of Lao PDR has been making major efforts in the past decade to mitigate economic barriers in accessing health services, including the launch of the National Health Insurance (NHI) scheme in 2016, and the consolidation of existing free maternal-child health services with other social protection measures for efficient management and higher risk-pooling [6].As a result, the country has achieved near-universal access to care, with 94% of the population covered by social protection schemes.The sector's financing, however, continues to depend on high levels of out-of-pocket (OOP) expenditures with households spending 45% of total health expenditures and decreasing public sector spending on health [7,8].
These challenges and an increasing burden of noncommunicable diseases are expected to result in increasing socioeconomic disparities in health.In this article, we use data from the Lao People's Voice Survey to study socioeconomic inequities across multiple domains of health system quality.

Setting
The Lao People's Democratic Republic (Lao PDR) is a lower-middle-income country in South-east Asia with a population of 6.5 million people [9].It is ethnically diverse with 49 ethnic population groups with the majority of the population (67%) living in rural areas [2,9].In 2018, approximately 18% were officially below the poverty line, defined as earning less than US$1.9 per day, which is among the highest poverty rates in the South-east Asia region [8].The COVID-19 pandemic is expected to have resulted in higher poverty rates and increased inequality with increasing fiscal pressure forcing the government to restrict health expenditure [8].
Healthcare in Lao PDR is largely delivered by the public sector with a small but growing number of private clinics and hospitals [10].In rural areas, government-managed health centres deliver primary care and refer people to district and provincial hospitals for complex health conditions.In urban areas, a growing number of private clinics and hospitals deliver care in addition to government hospitals [10].Government-managed health centres serve as the entry point for primary care but have low rates of utilisation because they are bypassed for better equipped and perceived higher quality hospital-based care [10].In Vientiane, the capital city, government managed specialty hospitals are often crowded due to bypassing of lower levels of care and paucity of public primary care providers within the city [10].

Data source
All data analysed was collected through the first Laos People's Voice Survey (PVS) conducted in 2022.The PVS survey tool is a cross-sectional instrument developed by the Quality Evidence and Health System Transformation (QuEST) network [11].PVS is designed to collect detailed health system performance measures including user experience, and perceived quality and confidence in the health system from the perspective of the adult (18 years old and higher) population.
The first PVS was deployed in Lao PDR between May and August, 2022 [12].It used age-, gender-, residence-and region-specific sampling targets to ensure nationally representative data.A total of 2007 interviews were completed using phones from a random sample of nationally active mobile users [12].This sample of completed interviews is consistent with national phone surveys such as Afrobarometer, Latinobarometer and, additionally, PVS undertaken in countries where phone ownership exceeds 80% [11,13,14].

Participants
Inclusion criteria were age 18 or older, and ability to complete the survey in Lao, Hmong or Khmou, which are the three most widely-spoken languages in Lao PDR.Respondents provided informed consent verbally before answering questions.

Measurement
We used health system measures in the study based on the framework proposed by Lancet Commission on High Quality Health System (HQSS), which focuses on outcomes and processes that matter most to people such as care experiences, perceived quality of care and trust in the system [15].In order to comprehensively assess socioeconomic differences in access to quality care, we analysed an extensive set of measures related to health system competence, user experience, perceived quality of care and confidence in the health system.Table 1 describes the outcome measures, their definitions, the questions asked to respondents, and how they were coded for analysis.We selected six binary measures to assess health system competence: whether respondents had a usual source or a regular provider for health care, whether their usual source for care was a government managed health centre (as opposed to any kind of hospital or private clinic), whether they received 3 or more out of 6 preventive health services in the past year, whether women received cervical screenings in the past year, whether women 50 years and older received mammograms in the past year and whether respondents had an instance of unmet need for healthcare in the past year.Only government-managed health facilities (health centres and hospitals) or licensed private facilities (clinics and hospitals) were considered as usual sources for health care; pharmacies and traditional practitioners were excluded.Given that most respondents seem to perceive quality of primary care to be higher in hospitals and private clinics, we interpret having public health centres as place of usual care as evidence of access to lower quality care.We included 6 preventive health services: blood pressure test, blood glucose test, blood cholesterol test, mental health check-up, vision check and teeth check-up.
We selected three binary measures to assess user experience and perceived quality among respondents that reported visiting a health facility in the past year: whether they had experienced discrimination in the past year, whether they experienced medical error in the past year, and their overall quality rating of their most recent visit to a health care facility.We used a binary definition for high-quality care; we considered care to be high-quality if respondents rated the overall quality of care received as 'Excellent' or 'Very good' out of 'Excellent', 'Very good', 'Good', 'Fair' and 'Poor'.
Only respondents that reported visiting a health facility in the previous year were included in the user experience and perceived quality of care measures.
We assessed confidence in the health system using confidence in their ability to receive quality health care when needed, confidence in their ability to receive quality health care and afford health care when needed.We considered respondents to be confident if they were 'Very confident' or 'Somewhat confident' out of 'Very confident', 'Somewhat confident', 'Not too confident', 'Not at all confident'.Measures of confidence were assessed among all respondents.

Explanatory variables
The main predictor of interest was poverty.Based on the latest data from Lao PDR, approximately 18% of households are classified as poor [2].To identify these poor households in the PVS data, we first ranked all households using household assets.During survey design, questions about 8 household assets were selected for inclusion in the survey based on their value in predicting household poverty levels in the 2017 MICS in Lao PDR [5].The survey instrument asked about household ownership of: a cement or ceramic roof, a clock, electricity, a bicycle, a motorcycle or scooter, a mobile phone, a computer, and a car.A principal component analysis was conducted using the asset questions, and the first principal component (describing 26% of overall variation) was extracted.Asset rankings were defined based on this first principal component, incorporating sampling weights.

Covariates
We adjusted for age and sex to isolate the association between socioeconomic status and health care quality independent of these factors.Age is defined as a categorical variable: 18-29, 30-49 and 50 and above and sex is defined as a binary variable (women or men).

Statistical methods
We used post-stratification weights to adjust raw data to reflect population demographic characteristics using an iterative proportional fitting algorithm known as raking [16].Population parameters were calculated using weighted data from the 2017 Multiple Indicator Cluster Survey (MICS) in Lao PDR [5].We started our analysis by describing the characteristics of the weighted sample.Next, we measured inequities in Lao PDR's health system by running unadjusted and adjusted logit models for each of the 12 health system performance measures described above.Finally, we introduced an interaction term between urban residence and poverty to study whether urban residence modifies the association between poverty and primary outcome measures.As a sensitivity analysis, we measured interactions using linear probability model for each outcome measure and provided results in the Appendix.

RESULTS
A total of 2007 completed interviews were analysed.Table 2 shows characteristics of respondents and quality measures, stratified by poverty status.Among the poor, a higher proportion were men (52%) compared to women (48%).The youngest age group between 18 and 29 years old accounted for 47% of the poor while the same age group accounted for 28% of the non-poor, and 4 in 5 (80%) of the poor were from the rural areas.Lower proportions of the poor had a usual source or a regular provider for care while a higher proportion considered health centres to be their regular provider.A lower proportion of the poor used at least three preventive health services (25%-36%) compared to the nonpoor population and only 9% of the poor women had cervical cancer screening tests compared to 17% of the non-poor.In contrast, a higher percentage (25%) of the poor women 50 years or older received mammogram tests compared to 13% of non-poor women.The poor in the sample reported higher (32%-27%) perceived quality of care of their most recent visit to a health care facility while reporting higher rates of discrimination (15%-11%) and medical error (9%-4%).The poor were less confident in their ability to receive quality care (72%-85%) and afford care (66%-85%) when needed.
Table 3 shows adjusted and unadjusted associations of 12 outcome measures with poverty.In adjusted models, poverty was associated with 55% (aOR 0.45 95% CI: 0.26-0.78)lower odds of having a regular provider for care, 56% (aOR 0.44 95% CI: 0.31-0.65)lower odds of receiving at least 3 preventive health services and 116% higher odds (aOR 2.16 95% CI: 1.35-3.48) of using a health centre as regular provider for care.There were no significant differences between the poor and non-poor in receiving cervical cancer screening tests (among women), mammogram screening (among 50 or older women), experiencing discrimination and medical error, perceiving quality of care in the most recent visit to a health facility, or having an episode of unmet need for care in the past year.
Poverty was associated with lower levels of confidence in ability to access and afford care when needed.In adjusted models, poor had 50% lower odds (aOR 0.50 95% CI: 0.34-0.72) of being confident in their ability to receive quality care, 50% lower odds (aOR 0.50 95% CI: 0.34-0.73) of being confident in their ability to afford quality care when needed.
Table 4 shows interactions between poverty and urban residence.There were similar associations for all outcome measure for the poor in urban and rural settings; none are significantly different.Linear probability model (see Additional File 1) also confirm these findings.

DISCUSSION
While LMICs have successfully expanded coverage of many essential health services, poor quality of care continues to cause a high morbidity and mortality burden, particularly among the poor, who often have less access to care [17][18][19] and receive substantially lower quality of care [20].Interventions that seek to address inequities have focused on improving access to care among the poor and other underserved populations [21].In general, less is known on inequities in quality of care; current evidence is mostly based on narrow patient-provider interactions and adherence to medical standards by the provider [20,22].
Recent studies using Service Provision Assessment (SPA) and Demographic and Health Survey (DHS) datasets, show access to quality care within countries is generally more restricted among the least educated, the poor and among populations with limited physical access to facilities [20].A study in Ethiopia found equitable access to poor quality of family planning and antenatal care across poor and rich sections of society [22].
In this article, we used a broader framework for quality proposed by the Lancet Commission on High Quality Health System (HQSS) that incorporates population perspectives on health system performance.We included measures beyond health service utilisation: competency of health systems to deliver preventive health services for the adult population and cancer screenings for women; unmet need for care; care experiences; perceived quality of care and confidence in people's ability to receive and afford good quality care when needed.More particularly, we used data from the PVS survey in Lao PDR to study health system inequities.
The results show significant inequities between the poor and non-poor.The poor have just half the odds of the non- poor of 'having a usual source of care' and of receiving preventive health services.When sick, the poor have statistically significantly lower confidence in their ability to receive and afford the care they need.The poor also have more than double the odds of using health centres as regular care providers.Among participants that visited a health facility in the past 12 months, the lowest proportion perceived the overall quality care in health centres as high-quality (See Additional File 2).While the non-poor tend to bypass health centres to seek care elsewhere, the poor continue to use care in these settings.While a high proportion (12%) of health system users reported experiencing discrimination or unfair practices during treatment and 5% reported experiencing medical error, the poor did not a have a statistically different odds of experiencing discrimination and medical error compared to the non-poor.Experiences of discrimination during treatment and medical errors have not been widely studied in Lao PDR.One qualitative study about quality of ANC care in public facilities found negative provider attitudes towards the poor and the use of informal payments, which can lead to providers favouring wealthier clients [23].In the PVS sample, discrimination may be occurring along different dimensions than poverty.
Confidence in health care or health systems is an important measure associated with care uptake and adherence to clinical recommendations [15].However, prior to our study, it had not been measured in Lao PDR.The study's findings of lower confidence among the poor to receive and afford quality care when needed is concerning.Using data from surveys in high-income countries, Bleich et al. found confidence or trust in health system was associated with measures outside of the health system or care experiences [24].Further research in low-income countries could explore the root cause of lower confidence in health system to explore opportunities to address it.Our findings of disparities in the coverage of health services are consistent with the most recent household Lao Social Indicator Survey (LSIS) survey [25].According to the LSIS, the poorest women were 45% less likely to have a skilled health personnel for ante-natal care (ANC), 50% less likely to have all three measures taken during ANC (blood pressured, urine sample and blood sample) and had nearly 3 times less likely to have their babies weighed at birth [25].While we found that poor and non-poor users reported similar levels of quality of care in their most recent visit to a health facility, disparities in the coverage of key preventive health services between the poor and non-poor likely contribute to disparities in health outcomes.
Our study has several limitations.While sample weighting should address selective responsiveness, it is possible that some groups are underrepresented in the PVS survey.Additionally, it is important to note that the survey was conducted during the COVID-19 pandemic; the observed utilisation patterns, expectations for quality care and confidence in the overall health system are all likely to have been impacted by the pandemic.Finally, the results were based on self-reports of use of services, care experiences and quality of care; no direct or objective measures of quality of care were collected.
Nonetheless, the study highlights substantial socioeconomic inequities in access to quality care despite government efforts to increase access to care over the last couple of decades.Additional effort to tailor universal health care policies to increase access to care for the poor is necessary along with strategies to improve their confidence in the health system, particularly in light to the increasing burden of non-communicable diseases.

CONCLUSIONS
The results presented in this article suggest that substantial socioeconomic disparities in access to quality care persist in Lao PDR despite major national efforts to provide universal access to care despite achieving near universal coverage of health insurance.The Ministry of Health may need a targeted approach to UHC that identifies and mitigates access barriers faced by the poor.This approach should also empower sub-national governments and health institutions to devise and implement locally-tailored approaches to meet the needs of the poor in local communities.
The implications of our findings extend beyond Lao PDR.A nuanced approach to UHC may be needed in other LMICs to reach the poor.In light of the diverse challenges faced by people in different regions and population groups, locally-customised strategies may be needed to ensure equitable access to quality health care.
Definitions of quality measures.
T A B L E 1 Coded 'Yes' as receiving the test and 'No' as not receiving the test Received mammogram tests (50+ women) Received mammogram screening in the past year Have you received mammogram tests?User experience and perceived quality of care Perceived the most recent visit to health facility as high quality Perceived most recent visit in the past year to a health facility to be high quality Thinking about the last visit: How would you rate the overall quality of care you received?Coded 'Excellent' or 'Very good' as high-quality; 'Good', 'Fair' and 'Poor' as non-high quality T A B L E 2 Sample description.
T A B L E 3 Adjusted and unadjusted associations of quality measures with poverty.Adjusted odds ratios for poverty and interaction of poverty and urban residence.
aThis table shows estimated associations between poverty and quality measures.Each coefficient corresponds to a separate logistic regression.b Covariates included sex (men, women) and age (18-29, 30-49, 50+).c Receiving mammogram was not adjusted because it only included women 50 or older.T A B L E 4 a Covariates included sex (male, female) and age (18-29, 30-49, 50+).b Covariates included age (18-29, 30-49, 50+).c Left blank because all 4 women (50 years or older) in the sample from a rural setting and poor did not receive mammogram tests.TROPICAL MEDICINE & INTERNATIONAL HEALTH