We found that recent strikes by health workers were associated with lower maternal health care utilization of ANC and delivery at a health facility among pregnant women in Trans Nzoia County in western Kenya. We also found evidence that strikes were associated with delayed child immunization for their first oral polio vaccine. Several recent studies have explored the impact of strikes by health workers on health services and outcomes in Kenya [4-9]. These studies mostly rely on hospital and inpatient data, and our study contributes important community-level data for outpatient and antenatal care to the literature. A study using hospital admissions data at 13 County referral hospitals during the 2016-2017 nationwide physicians’ and nurses’ strikes found that inpatient admissions decreased significantly during strike months, with the authors estimating that 183,170 individuals were expected but did not receive inpatient care at these hospitals over the course of the strikes, including 60,000 maternity patients . Another study using records from 18 County referral hospitals and 14 faith-based health facilities found that the proportion of fully immunized infants fell 57% in public facilities and increased 252% in faith-based ones during the 2017 nurses’ strike . In our study, a lower proportion of children received their first oral polio vaccine in the strike group (83% versus 89%) but this was not statistically significant, although children in the strike group did receive their vaccine significantly later after birth.
While our study was not designed to detect differences in maternal or child mortality, studies reported mixed findings on the association between strikes by health workers and mortality in Kenya [6, 7]. In the only study to date to use population-level health outcomes data, during six strikes by health workers ranging from 9 to 42 days between 2010 and 2016 in Kilifi County, Kenya there was no change in all-cause or cause-specific mortality during strike versus non-strike days . The authors suggest a number of potential explanations for their findings including service delivery continuing in private and faith-based facilities, relatively short duration of strikes, decreased exposure to poor quality inpatient care and high-risk procedures, and the fact that a large proportion of deaths (during strike and non-strike periods) occur outside of health facilities. Population-level data on maternal and child health utilization and outcomes during the study period were not publicly available at the time of writing, but these data will provide important additional insights into the impact of strikes in 2017.
Given that the strikes in 2017 lasted longer than previous strikes in Kenya, it is possible that they were more likely to affect population-level mortality that is not captured by studies using inpatient data [28, 29]. Decreased ANC coverage during strikes could have also interrupted prevention of mother-to-child transmission services for women living with HIV, but there are no published studies investigating strike-related changes in mother-to-child transmission of HIV. However, even in the control group, our study found that only 70% of participants attend 4 or more ANC visits and only 73% delivered at a health facility with a skilled health worker when there were no health worker strikes. Several studies have suggested that the poor quality of facility-based maternal and child health care may account for slower reductions in maternal and neonatal mortality compared to increases in service utilization and facility-based care [11, 30-32]. Research also shows that perceptions of poor quality maternal and child health services among mothers, including long wait times, lack of providers and essential equipment and drugs, disrespectful care, and out-of-pocket payments, represent important barriers to seeking care . Many of these issues related to poor quality of care and ill-equipped providers and facilities were at the center of debates during recent strikes, but it is unclear whether strikes have led to improvements in health care services for patients.
There are several limitations to this study. First, we relied on retrospective data from Mother-Baby Booklets and self-report to measure maternal and child utilization of services and outcomes. There was significant missing data for one of our outcomes – oral polio vaccine – so these results should be treated with caution, but there was also the potential for recall bias for women who self-reported outcomes. We did not find evidence of increased pregnancy complications or child mortality, however, 37% of women from the parent study strike group were not able to be re-consented. The only demographic data available on missing women was age, which did not differ between missing and re-enrolled women. We were also not able to measure maternal mortality and it is possible that some women who were not able to be relocated had in fact died or had other complications.
Second, we present data from only one County in Kenya. While our findings are generally consistent with the literature available from other parts of Kenya, it is possible that maternal and child health services and outcomes were affected differently by strikes in other parts of Kenya. Trans Nzoia is poorer, more rural, and has generally worse maternal and child health indicators by comparison to Kenya overall, however, it does have a mostly representative split between public and private health facilities . Third, we did not assess the impact of a specific health worker strike. Most participants in the strike group delivered during a nationwide nurses’ strike but were also pregnant at some point during the physicians’ strike. Moreover, it is common in health facilities when one cadre of health worker is on strike, services may be suspended, and other cadres of health workers may not be working. Thus, it is difficult to assess exactly what health facilities and services remained functional during 2017.
Finally, there may have been changes in the health system outside of the strike that affected maternal child health services and utilization from 2017 to 2018 that were difficult to account for in this study. In our analysis, we controlled for enrollment in the Linda Mama health insurance program that was officially launched in 2017. In our study population, only 15% of women had health insurance at the time of delivery in 2017, but 59% had health insurance at the time of delivery in 2018. While enrollment in Linda Mama allows women to access maternity services for free in accredited private sector facilities that mostly remained open during recent strikes in the public sector, it is unclear whether higher rates of enrollment would have led to greater access to care in 2017 given that many private sector facilities were quickly overwhelmed by demand .