Health care seeking in modern urban LMIC settings: evidence from Lusaka, Zambia

Despite significant improvements over the past 30 years, mortality rates in LMICs remain high: 4% of children in LMICs die before their 5^th birthday, and preventable mortality from both infectious and chronic conditions is significantly higher than in high-income countries [1, 2]. Many efforts to improve health outcomes in LMICs have focused on improving access to primary health care services through interventions such as the removal of user fees for services in public primary health facilities [3‐8]. However, there is widespread evidence that the average quality of care provided in health facilities in many LMICs is low [9‐16]. In addition, quality of care tends to vary significantly across health facilities, creating a complex decision-making environment for patients who seek care [17‐19].

There is growing evidence that patients in LMICs are increasingly aware of differences in quality of care, and often bypass primary health facilities in their communities to seek care at more distant or higher-level health facilities [20]. Extensive bypassing has been documented for childbirth [21‐28]: for example, in a study in Uganda, 29% of women bypassed their nearest health facility for delivery [25]; in a study in Nepal, 71% of women whose nearest facility was a birthing center bypassed the center to deliver in a hospital [24]. Studies have also documented high rates of bypassing for primary care in settings such as China, Ghana, India, and Chad [29‐32], and for inpatient care in Sierra Leone and Kenya [33, 34]. Fewer studies have examined bypassing for pediatric care [34‐36], but these studies also show high rates of bypassing. Important predictors of bypassing include distance to a hospital [28] and perceived quality of the local primary health facility [22, 32, 37]. Bypassing in urban areas, where patients have more options for where to seek care and their choices are less constrained by distance, may be particularly revealing of patient preferences. While evidence on bypassing patterns in urban areas is relatively scarce, the existing evidence suggests that there are often higher rates of hospital use in urban areas relative to rural areas [31, 35, 38].

In this study, we describe care-seeking patterns among urban informal sector households in Lusaka, Zambia. Thanks to a 2012 reform [6] patients in Lusaka are not required to pay fees for primary care as long as they access care through health posts or health centers. Despite these financial incentives to use lower level facilities, there is evidence that many families bypass local health centers and directly seek care either at hospitals or in the private sector [39].

To assess the extent of bypassing, we collected detailed treatment seeking data from 620 randomly-selected households in Lusaka, and identified the location and type of facilities used for adult as well as child healthcare. We quantify the rates of three types of bypassing: i) not using health centers or health posts for primary care (non-compliance with government recommendations); ii) seeking care outside of the residential neighborhood (spatial bypassing to reach higher quality facilities), and iii) directly seeking care at tertiary teaching hospitals (bypassing two levels of care).

A total of 753 randomly selected households were approached by the study team. Nine households (1.2%) were excluded because the respondent was above 65, 43 households (5.7%) could not be reached and 26 (3.5%) indicated they were too busy or not interested in the study. Forty-eight households (6.4%) were employed in the formal sector, and also excluded from the study. We therefore interviewed 627 adults about their recent care-seeking behavior and that of children in their household. Three EAs had less than four eligible households due to high formal sector employment in these neighborhoods – we excluded households from these areas from the analysis (N = 7, 0.9%) because the number of observations was too small to establish the most commonly used health facilities in these settings. A sample flow diagram is included in Additional file 1: Figure S1.

The final adult analytic sample included the 577 adults whose most recent visit to a health facility was for non-emergent care. The majority (78%) of participants were female (Table 2). About one quarter (24%) of the sample was over age 45, 43% was aged 30–44, and 29% was under age 30. The majority (59%) of the sample had completed secondary education or higher. The most common reason for their most recent health visit were new health problems (54%), followed by routine check-up (24%), and chronic disease treatment (22%). On average, the households in the sample had two general hospitals, 16 private facilities, and 11 other health facilities within five kilometers of their homes.

The survey participants had a total of 402 children under-5 living in their households, of whom 141 had sought care for an episode of diarrhea (63%), fever (46%), cough (67%), or fast breathing (10%) in the past two weeks. About half (49%) of these 141 children were female.

Figure 1 shows the spatial location of all health facilities officially recognized by the Ministry of Health within the District of Lusaka. There were a total of 88 facilities operating in Lusaka district based on the list from the Ministry of Health: two teaching hospitals, six general (Level 1) hospitals, two Level 2 hospitals, 47 private facilities and 31 smaller facilities, including health centres, health posts or mission facilities.

Figure 2 illustrates the distribution of facilities used for care by reason for seeking care. Across all care or health problem categories, Level 1 hospitals were the most commonly used facility type, with less than one third of adult patients using health posts or health centers for checkup, chronic or acute care. Among adults, non-governmental facilities (private or faith based) were most commonly used for check–ups (11%) and teaching hospitals were most commonly used for chronic care (18%). Compared with adults, children were more likely to receive care in a health post or health center (with 41% seeking care at these facilities), or a pharmacy or drug shop (21%). One third of children received care in a hospital.

As shown in Table 3, bypassing was very common across all conditions: on average 71% (95% CI: 67% to 75%) of adults bypassed public health centres and posts, with particularly high rates for chronic conditions (77%; 95% CI: 70% to 85%). Horizontal bypassing was less common: 32% (95% CI: 29% to 36%) of adults visited a more distant rather than a nearby health facility, and this rate was similar across different reasons for health visits. Finally, the rate of two-level bypassing among adults was 8% (95% CI: 6% to 11%), with the highest observed rate for adults seeking care for chronic conditions (18%; 95% CI: 11% to 25%).

The primary care bypassing rate among children was 59% (95% CI: 51% to 67%), slightly lower than the rate among adults. The bypassing rate was similar for children with different symptoms. The rate of horizontal bypassing was slightly higher among children than among adults at 45% (95% CI: 37% to 54%). Among children who bypassed the nearest health facility, 47% (95% CI: 35% to 59%) went to pharmacies and the remainder sought care at more distant public primary care facilities or hospitals. Finally, the rate of two-level bypassing among children was 1% (95% CI: 0% to 2%).

Figure 3 illustrates the spatial patterns of bypassing. About two thirds (67%) of the overall primary care bypassing occurs at local (Level 1) hospitals, which are located within the same constituency and thus are within two km of most households in our sample (Fig. 3, Panel A). Horizontal bypassing involves on average slightly larger distances (Fig. 3, Panel B). About half of horizontal bypassing goes to hospitals in other constituencies (UTH and Matero Level 1 hospital appears to be most popular in our sample, accounting for 20 and 14% of total horizontal bypassing, respectively) – the rest of the patients seek care at a mix of public (30%) and private or other facilities (19%) in other parts of the city. Distance travelled is on average largest for two-level bypassing, and mostly concentrated at the University Teaching Hospital (UTH) (Fig. 3, Panel C), which attracts patients from the entire city.

Bypassing rates varied significantly across the different constituencies in the sample (Additional file 1: Table S1). The rate of primary care bypassing ranged from 28 to 100%, the rate of horizontal bypassing ranged from 5 to 79%, and the rate of two-level bypassing ranged from 0 to 32% across constituencies. The large differences in care seeking behavior can be best illustrated by comparing two constituencies with very different behaviors: in one EA in Lusaka Central near Bauleni Health Centre, only 10% engaged in primary care bypassing, 15% in horizontal bypassing, and only 5% went to teaching hospitals (two-level bypassing). In contrast, in another EA near Chilenje Level 1 Hospital, the rates of bypassing were 95% (primary care bypassing), 47% (horizontal bypassing), and 32% (two-level bypassing).

As shown in Table 4 and Additional file 1: Table S2, bypassing rates varied with respondent characteristics. Among adults (Table 4), women had a 10% lower odds of primary care bypassing (95% CI: 0.83 to 0.98) and a 10% higher odds of horizontal bypassing (95% CI: 1.00 to 1.20) than men, after adjusting for other characteristics. Married participants had a 10% lower odds of horizontal bypassing (95% CI: 0.84 to 0.98) than unmarried participants, though rates of primary care bypassing and two-level bypassing were very similar between married and unmarried participants. Older respondents had higher rates of two-level bypassing and horizontal bypassing, though these associations were only statistically significant for two-level bypassing. Adults with a higher socioeconomic status as measured by education level and asset scores generally had higher rates of bypassing than those with lower socioeconomic status, though this association was not statistically significant for all outcomes and education levels. The finding (from unadjusted analyses) that two-level bypassing is more common among adults seeking care for chronic conditions than other types of care persisted after adjustment for socioeconomic characteristics.

Among children (Additional file 1: Table S2), primary care bypassing was higher among those whose caregivers had completed secondary education than those with primary education or less (odds ratio 1.27, 95% CI: 1.06 to 1.53), but there were no statistically significant differences by education level for two-level bypassing or horizontal bypassing. Bypassing rates also did not differ significantly by the asset quintile of the caregiver, after adjusting for other characteristics. Primary care bypassing was significantly less common for female children (odds ratio 0.78, 95% CI: 0.66 to 0.92) than male children, but other forms of bypassing did not vary significantly by gender. Bypassing rates were generally lower among children presenting with fever and higher among children presenting with diarrhea or fast breathing, though these associations were generally not statistically significant.

In this study, we described care-seeking patterns in Lusaka, Zambia and measured the rates of primary care bypassing, horizontal bypassing, and two-level bypassing. Despite recent government efforts to encourage use of primary care through the removal of user fees, primary care bypassing is extremely common in Lusaka, and Level 1 and Level 3 hospitals are used extensively for non-emergency care. These findings are consistent with a growing literature showing high rates of bypassing in low- and middle-income countries [20‐34, 36, 37, 45‐48]. Our study builds on the existing literature by mapping bypassing patterns in an urban setting. In the context of rapid urbanization in sub-Saharan Africa, where the proportion of the population living in an urban area increased from 27 to 41% over the past 30 years [1], it is important to understand care-seeking patterns in cities. Furthermore, while past studies tended to focus on a single definition of bypassing, we examined the rates of different forms of bypassing and are thus able to further understand different care-seeking patterns. While we found very high rates of primary care bypassing (71% of adults and 59% of children), we found lower rates of horizontal bypassing (26% of adults and 45% of children).

High rates of hospital use for non-emergency care, as observed in this study and others [35, 49], present a challenge for achieving the Sustainable Development Goal for universal health coverage [50]. The World Health Organization (WHO) has called for a shift of the entry point to the health system from hospitals to primary care centers to promote efficient use of resources, equitable access to care, and continuity of care [51]. In Zambia, the user fee structure is set up to discourage the use of hospitals as a first point-of-contact. While hospitals could attempt to stop this practice, it is possible that the bypassing fee incentivizes them to accept patients seeking non-emergency care.

The extensive use of pharmacies and drug shops for pediatric health care observed in this study also presents a potential challenge. Pharmacies play a significant role in primary care provision in many LMICs, often because they are considered to be convenient locations to seek care [52, 53]. However, there is evidence of important gaps in pharmacists’ education and training in many settings [52, 54], and pharmacies often lack basic medications and equipment for primary care provision [54]. Furthermore, a study in Zambia found widespread non-prescription sale of antibiotics in community pharmacies, a practice that may contribute to antimicrobial resistance [55]. It is important to understand why caregivers choose to bring their children to pharmacies instead of free public facilities. If pharmacies are to continue playing an important role in pediatric care in Zambia, there is a need to ensure that they are adequately staffed and supplied, and that measures are in place to ensure appropriate use of medication in these locations.

While this is an observational study and does not provide direct insights into reasons for bypassing, our analysis and the existing literature point to several possible explanations. First, patients may bypass because they perceive care to be of higher quality at a more distant or higher-level facility [22, 37]. In our data, these perceptions seem to vary substantially across communities: in some EAs, nearly all patients bypassed the local primary care facility while, in others, it was much more commonly used. Higher-income patients, in particular, may be willing to pay more to receive care that they perceive to be of a higher quality [29, 32, 35]; this may help explain our finding that bypassing is more common among study participants with higher levels of education and household assets. A second possible explanation is that the hours of operation of the bypassed facilities are too limited or inconvenient [56, 57], leading patients to seek care in facilities with hours that are more amenable to their schedules. Another possible explanation is that patients bypass nearby facilities due to fear of stigma from seeking care in their own communities for conditions such as HIV/AIDS. In our analysis of horizontal bypassing, we found that some patients bypassed nearby primary health centers to seek care at more distant primary health centers, while other patients bypassed nearby hospitals to seek care at more distant hospitals. The estimated HIV rate in Lusaka is 16% [58], and care-seeking for HIV/AIDS is associated with high levels of stigma [59]. Past studies in LMIC settings have found that patients may be willing to travel longer distances to avoid being recognized when seeking testing or treatment for HIV/AIDS [60, 61], so it is possible that participants in our study chose to bypass nearby facilities for this reason. Finally, many hospitals in Lusaka were upgraded from health centers in recent years [62]; it is possible that residents were unaware that they were using hospitals, though the fee structure would likely make it clear. This is an important area for future research.

The strengths of this study include the use of a dataset with a complete mapping of facilities in a major urban center that is likely representative of many urban areas in sub-Saharan Africa, and the detailed data on care-seeking behavior collected from a randomly selected household sample. These descriptive data can be used by local managers to inform analyses of bypassing behaviors and subsequently consider how to address them.

This study also has several weaknesses. First, we do not have information on whether bypassing patients were referred to higher level facilities by providers in primary health facilities, or were attending follow-up visits which can occur in specialized clinics in teaching hospitals. These care-seeking patterns would be in line with Ministry of Health guidance. While referrals and follow-up visits might help to explain the high rates of two-level bypassing by patients with chronic conditions (as 18% of patients with such conditions seek care at UTH), they are unlikely to explain the broader trends we observe in this study since we found that patients seeking care for new health conditions bypassed at only slightly lower rates than those seeking care for chronic conditions. Data on referral patterns – including whether patients were referred from primary care to higher level facilities, sought care at primary care facilities before deciding themselves to go to higher level facilities, or went straight to higher-level facilities – would help to shed further light on the challenges at the level of primary care facilities. Second, our household survey included informal sector households only. However, this is the large majority of residents in Zambia [41], and only 6.4% of the adults we approached for the study were employed in the formal sector and excluded for this reason. It seems unlikely that bypassing behavior would be less pronounced in the formal sector given the generally higher socioeconomic status of these households – assessing these differences would certainly be interesting for future studies. Third, the structure of hospital services in Zambia will be updated in 2022 as part of the 2022–2026 National Health Strategic Plan. However, the hospital mapping we used in this analysis was current for the study period and the Ministry of Health’s guidance regarding the use of primary care is not expected to change. Fourth, the time horizons are different for the child sample (past two weeks only) and the adult sample (most recent visit); this may impact our comparisons between adults and children. Finally, we make the assumption that individuals were living in their current household and were at home when they most recently sought care. If many individuals moved between when they sought care and when they were interviewed, or if they sought care during their working day, this might change our results for horizontal bypassing; however, it would not change our results for primary care bypassing or two-level bypassing.

The results presented in this paper suggest that bypassing is incredibly common in Lusaka, and that existing care-seeking recommendations by the government are largely ignored. As policymakers aim to encourage the use of primary care, it is important to consider how to make lower level facilities more attractive and beneficial to patients. Hospital fee structures such as the one introduced in Zambia, whereby patients can access free primary care but have to pay to directly access care at a hospital, do not seem to deter patients from seeking care in hospitals; this suggests that patients highly value the care provided in hospitals.