Assessment of Bypass of the Nearest Primary Health Care Facility Among Women in Ghana

Key Points Question How frequently do women in Ghana bypass the health care facility nearest their homes in favor of more distant facilities perceived to offer better care, and what factors are associated with this choice? Findings In this survey study including 1946 reproductive-aged (ie, 16-49 years) women in Ghana in 2017, 32% of women reported bypassing their nearest health care facility. Women who bypassed sought care at hospitals and private facilities more frequently and paid nearly 2-fold in out-of-pocket costs for care compared with women who did not bypass. Meaning These findings suggest that the services offered at many primary care facilities in Ghana may not be meeting the needs of women.


Introduction
Recent global initiatives have embraced ambitious new goals to improve primary health care service delivery and achieve universal health coverage. Three 2018 reports [1][2][3] have highlighted that significant work remains to be done to measure and improve quality and that merely expanding access to care is insufficient if the care provided does not meet patients' needs.
Evidence shows that local health care facilities are frequently not used or are bypassed by individuals in favor of secondary or tertiary-level hospital facilities even when care could be managed at the local primary health care level. [4][5][6][7][8] This underutilization of dedicated primary care facilities may lead to increased travel and cost burdens for patients, increased costs to health systems through inefficient utilization, overburdening of higher level facilities, and subversion of the goals of universal health coverage. 4,[9][10][11] The prevalence of bypassing varies widely within and between countries and health systems. 2, 4,6,[8][9][10][11][12][13][14][15][16][17][18][19] Many individuals bypass what they perceive to be understaffed and poorly equipped public government-run health facilities to seek care in private facilities and hospitals. 10 Several studies have described substantial variation in individual behavior across different health systems, particularly with regard to childbirth facilities. [4][5][6]11,14,17,[20][21][22][23][24][25][26][27][28][29][30][31][32][33] There is limited evidence characterizing the extent of primary health care bypass in Ghana and West Africa in general. A better understanding of these bypass behaviors may provide insights into local primary health care improvement efforts for policy makers and facility managers. Recent surveys by Performance Monitoring and Accountability 2020 (PMA2020) 34,35 provide nationallyrepresentative data on care seeking and utilization of primary health care in Ghana. Using these data,

Study Area and Data Collection
Extended description of the data collection methods used by PMA2020 can be found elsewhere. 34,35 The survey was designed to provide nationally representative data on key health and languages. To obtain the data, the Ghana Statistical Service selected 100 enumeration areas across all 10 regions with probability proportional to size using a master sampling frame stratified by urbanrural areas. Within these enumeration areas, 42 households were selected using a random start method to complete the household survey, and all members of the household were surveyed. All data used in this study were collected during round 6 of the PMA2020 survey conducted in late 2017.

Bypass Definition
In this study, bypass was defined as a woman's report that she sought care at a health facility other than the nearest facility to her residence ("Is this facility the closest facility to your place of residence?" with binary yes or no answer). The woman was then categorized as having bypassed her nearest facility or not.

Variables
We

Statistical Analysis
All women who responded to the survey and had visited a health facility in the last 6 months were included in the analyses. To generate nationally representative, population-based estimates, we used survey-weighted summary statistics that accounted for the multistage clustered survey design in all analyses.
We used Poisson regression models with a log-link and robust SEs to estimate the associations of bypass (main variable) with women's self-reported experience of their care. Outcomes were dichotomized as highest or most positive ratings vs all other ratings combined (a top-box categorization). The exponentiated regression coefficients from the Poisson regression can be interpreted as relative risks (RRs) comparing the proportion of those giving the highest rating between women who bypassed and those who did not. We used linear regression with robust SEs to estimate mean out-of-pocket costs paid by women who bypassed and those who did not. All models were first fit unadjusted for other variables, and then with adjustment for demographic and geographic factors and reasons for seeking care (eAppendix in the Supplement

Results
A total of 4289 women met eligibility criteria, and 4207 women (98.1%) were available, consented to the interview, and completed the interview. All 4207 women were asked whether they had visited women who had sought care in the last 6 months and lived in the sample household, which was 1946 women after reweighting to account for the survey design (eFigure in the Supplement). Table 1 provides summary statistics for the sample, stratified by bypass status. Using methods accounting for the multistage sampling design and rounded to the nearest integer, 629 women (32.3%) reported bypassing their nearest facility to seek care.

Demographic Characteristics Among Women Who Bypassed
A total of 305 of 975 rural women (31.3%) bypassed their nearest health facility, compared to 324 of 971 urban women (33.3%) ( Table 1). Women in urban areas tended to be wealthier than those in rural areas, with 335 women (34.5%) in the highest wealth quintile compared with 23 women (2.3%) in rural areas (eTable 1 in the Supplement). There were few women in the highest quintile of wealth in rural areas; however, these women had higher levels of bypassing than the wealthiest women in urban areas (eTable 1 in the Supplement).
Women who bypassed their nearest facility were more likely to be seeking care for themselves (as opposed to for a child or other family member or friend) than women who did not bypass (465 ; however, overall reasons for care-seeking were generally similar between women who bypassed and those who did not ( Table 2), suggesting that differing clinical needs were not the major driver of bypass. In particular, the proportions of women who sought care owing to a community health worker referral were similar among women who did not bypass and those who did (13 women [1.0%] vs 4 women [0.7%]). Some women may have self-referred to a more distant facility for strategic reasons. To further examine these patterns, we performed a sensitivity analysis in which we examined strategic bypass. This analysis included only women who bypassed their nearest facility because it was closed or did not offer their desired services. After looking only at this selected group of women, we found no significant differences in our primary findings (eTable 2 in the Supplement).
Rural women who bypassed typically did not seek care at community-based health planning and services facilities (eTable 1 in the Supplement). Instead, women who bypassed frequently sought care at hospitals or polyclinics and private facilities. Wealthy rural women were also much more likely to bypass their nearest facility than their urban counterparts (eTable 1 in the Supplement).

Self-reported Experience of Women Who Bypassed vs Those Who Did Not
Women who bypassed generally reported receiving more responsive care (

Discussion
This survey study used a nationally representative survey to estimate the countrywide prevalence of health facility bypass in Ghana. Unlike previous studies that focused on childbirth or emergency services, our analysis focused on a general population's experience of primary care and offers insights into who bypasses, why, and the outcomes associated with their care. We found that bypassing was relatively common: 32.3% of women reported bypassing. Moreover, bypassing was mostly owing to  the knowledge that the desired services were not available at the closest facility. Women who bypassed cited clinician competence as the most important factor in choosing a facility more frequently than women who did not bypass, indicating that women highly value competent clinicians and are willing to travel farther to get them. In rural areas, wealthier women tended to bypass more frequently, suggesting differences in behavior and functional access between socioeconomic strata.  Women who bypassed paid more out of pocket for their care than women who did not bypass.

JAMA Network Open | Health Policy
To our knowledge, this is the first study to examine costs associated with bypass in this context.
Other studies have examined discrete choice experiments of health facility utilization and quality and found that patients are generally willing to pay more for higher-quality services, although how much more is greatly dependent on context. 20, 21 Kahabuka et al 7 found that caretakers in Tanzania frequently bypassed nearest facilities for their children owing to lack of diagnostic equipment, insufficient drug stocks, limited hours, and poor services-similar to our findings. However, comparisons of bypass across countries can vary widely depending on the service being sought and the distribution of facilities offering that service. For example, in a 2020 study of family planning in Tanzania, 37 67% of women bypassed their nearest facility to seek contraception. The nearest facility   was a mean of 1.2 km away, while the chosen facility was a mean of 2.9 km away. In contrast, a 2016 study of utilization of childbirth facilities in Ghana 38 found that women lived a median 3.3 km from their nearest birth facility, with the nearest hospital 13.9 km away (bypass was not assessed). These differences reflect heterogeneities in opportunity for choice of facility, proximity of facilities offering a desired service, and burden associated with bypassing the nearest facility. Our population of women was also fairly young and as such, less likely to be seeking care for the types of chronic conditions and serious illnesses that might be associated with referral to hospitals.
Women generally reported similar reasons for seeking care whether they bypassed their nearest facility or not. This may suggest that many women's reasons for bypassing were less associated with the nature of their underlying health concerns and more associated with their perception of quality of care or experiences at the facility. There are limited published data from other low-and middleincome countries on differences between individuals who bypass and those who do not in reasons why individuals might seek care. A 2011 study by Gauthier et al 11 examined reasons for seeking care by facility type in Chad and reported that malaria and diarrhea were the primary reasons for seeking care; however, the study did not assess whether these reasons differed by bypass status.
Previous studies 4,6,[9][10][11]18,19,39 have suggested that bypass is frequently driven by perceptions of the quality of care offered. In our study, women who bypassed were more likely to rate their clinician choice, cleanliness of the facility, and privacy as excellent. However, despite rating these aspects of their facility highly, women who bypassed were no more likely to report excellent ratings for the overall quality of their care or to recommend the facility highly to others. This finding may be associated with differential expectations of what is considered to be acceptable care, with individuals who bypass having higher expectations than whose who do not. 40 In previous studies, 18,41,42 vignettes describing aspects of care indicated that individuals have wide variation in what they consider to be acceptable care, particularly those who are poorer and less educated.
Taken in the context of a growing emphasis on quality in health care delivery, our results have important policy and public health implications. They suggest that bypassing is common, expensive, and frequently associated with perceptions that the closest facilities do not provide the desired services. Women frequently reported that the competence of their clinician was of highest importance, and were most likely to bypass community-based health planning and services facilities, which do not have physicians on staff. Increased staff training, higher quality of care, and range and readiness of available services may decrease bypassing, reducing costs to patients. 7 Further research is necessary to identify which specific services can be brought closer to patients, if possible. Our results also indicate an opportunity to improve the system's efficiency in being a point of first contact for patients, per Starfield's four pillars of primary care 43 (ie, first contact, continuous care, comprehensive care, and coordinated care). While further research in this area is essential, improving the quality of care at local levels may help toward building more equitable, higher quality primary health care systems that better address patients' needs. 44,45

Limitations
Our study has limitations. First, this is a cross-sectional survey designed to be nationally representative of reproductive-aged women in Ghana, so our findings should be interpreted with caution and may not be generalizable to other subpopulations. Second, our definition of bypass was based on women's self-report of not visiting their nearest facility, which may be inaccurate, as many women may have been unaware of precise distances between their home and facilities. However, this definition has been used in previous studies, and our assumption is that women's perception of distance, rather than actual distance, is likely to be more relevant to their choice of health care facility. 6,9,19,30 Third, although our study focused on women who visited a facility, some women reported that they were unable to visit a health facility altogether, despite a desire for care. Fourth, because our study collected only limited data on reasons why women may have bypassed their closest facility, care should be taken in drawing definitive policy and public health conclusions based on our findings alone.