On November 12th in the Indian state of Chhattisgarh, twelve women who had received tubal ligations died. The tragic incident highlights the unfortunate reality that for many people around the world, hospitals and clinics may not satisfy the most basic assumption that visiting them will make you better. Equally worrying is the Indian government’s singular focus on increasing ‘institutional deliveries’ and family planning that led it to celebrate a surgeon who had performed 100,000 sterilizations, now spending no more than 4 minutes on each “case”.
Around 9 am one of us (Das) arrives at the Community Health Center in a small town in Madhya Pradesh, India. There is a row of women lying on the floor outside the labor and delivery room, moaning or groaning in various stages of pain and clear discomfort. We pay them no heed; this is a sight we have seen before. Over two years, we have visited many health facilities in the state and they are either (i) deserted or (ii) packed to the brim with women giving birth, about to give birth, or just having given birth. Mass tubal ligations are performed on as many women as will consent—aided by the promise of money under India’s conditional cash transfer for institutional delivery—which, till recently would not pay women who were giving birth to a third or higher parity child unless they also consented to be sterilized. The obstetrician/gynecologist is happy to show us around. She is young and clearly exhausted. “I came here 5 months back. Since then, I have been alone with one nurse in this facility. Every month we do 1,000 deliveries and around 500-600 tubal ligations; almost 100 are C-sections. I am here 18 hours a day and even then at 2 every morning I stay up bleaching the operating room.” She shows me the bleach and the OR is truly spotless. She tells me that she is looking for the first opportunity to leave—either by pursuing a specialization or leaving public service. “This is no way to help. I thought I would work on improving people’s lives. But there is no help from the government and I can’t keep this up much longer”.
A short 30-minute drive away, we visit the nearest Primary Health Care center. A young man is waiting next to a motorcycle. He has been crying. “I brought my wife here to give birth last night. But when the child was born, the doctor said that he did not have sufficient oxygen. He told us to take the child to the district hospital. I rushed my child there on the motorcycle, but by the time a doctor saw the child it was too late. No one cared. This doctor is not bothered. If only he had worried a little. He just did not care.” He keeps repeating the same phrase again and again: “No one cared” and it remains with us for a while.
Spurred by an international consensus that the way to meet the Millennium Development Goals on reductions in maternal, child and infant mortality was to increase the number of institutional deliveries, countries have devoted themselves with a single-minded focus to doing just that. The Indian Government, encouraged by the “success” of the Chiranjeevi voucher scheme , expanded conditional cash transfers for institutional deliveries to the entire country. In Rwanda, facilities were paid for institutional deliveries. The results have been impressive. Between 2005 and 2011 (the last year for which public data are available in India), institutional deliveries increased from 20% to 49% in nine Indian states with relatively worse socioeconomic and health indicators. Similarly, between 2000 and 2008 in Rwanda, institutional deliveries increased from (approximately) 25% to 75%. What has the fast rise in institutional deliveries done to child mortality?
Figure 1, based on ongoing work with Shareen Joshi and Quy-Toan Do shows the pattern of U5M and infant mortality in India since 1980. Under-5 mortality has fallen sharply over this time, while infant mortality has been harder to budge. Consequently, the share of deaths due to infant mortality has increased dramatically, and by 2008, accounted for more than 80% of all U5 deaths. Second, infant mortality was declining somewhat till 2004. After that there is a sudden drop in 2005, followed by a flat-line. Between 2005 and 2008, infant mortality budges not an inch. We look at that sudden drop and show that it can be entirely accounted for by differences in survey methodology between the 2004 and the 2008 surveys. Using institutional variation in eligibility, we confirm findings from other research that the CCT has had no discernible impact on infant mortality. If anything, the rise of institutional deliveries is remarkably consistent with the halting of a slow decline in infant mortality from the 1970s onwards.
Avid followers of this topic will point to Lim and others study in the Lancet with the opposite results. Unfortunately, the Lim study is fatally flawed as shown by Mazumdar, Mills and Jackson and by Randive, Diwan and Costa. India observers will also point to summary reports from the subsequent annual health surveys in 9 states showing that infant mortality has declined since then. These data have not been publicly released for research and replication, and our work on the previous surveys shows that replication and consensus among multiple teams is critical to a shared understanding of the data.
But it gets worse. It turns out that even the “successful” program on which the entire CCT was premised had no impact whatsoever on anything at all. The COHESIVE group, which works on healthcare in India, went back to the state of Gujarat that started and implemented the Chiranjeevi scheme and conducted a large household survey to understand the impacts of the program. Their results show that the program was ineffectual. Of course, by the time their evaluation appeared, success had long since been claimed, both nationally and internationally. The Singapore Economic Development Board and The Wall Street Journal had already decided in 2006 that the program had “drastically reduced maternal and infant deaths” and therefore deserved an Asian Innovation Award.
”Wait,” experts will say, “Indian health care is really poor. It should be patently obvious that increasing institutional deliveries will decrease infant mortality in countries where healthcare works.” But there is no evidence for that either. A recent study from Rwanda shows that their pay-for-performance program led to a sharp increase in institutional deliveries, but there was no impact on neonatal mortality. In any case, recommendations to increase institutional deliveries never add the caveat “where facilities are working properly” or provide concrete advice on how to make them work before all these extra institutional deliveries are performed. We might all agree that “if things worked well then there should be more facility based births.” But if things do not work well, do we say there should be more such births anyway? And if things worked well, do we know that people would not already be using the institutions to deliver children in a safe environment?
Worse, the policy to increase institutional deliveries is likely to make it harder to “make them work properly”. In Chennai public hospitals, the Caesarean section rate is over 60%. The enormous traffic makes scheduling deliveries essential. Predictable Caesareans become more attractive to the facility leading to higher-risk deliveries and rates of infection. In other parts of the country, Caesarians being “more attractive” takes on darker meaning as ongoing research finds medical officers in public hospitals carrying on a lucrative trade performing such operations on women waiting for natural delivery. Combined with the customary tubal ligations that improve the doctor’s statistics, this is doubly profitable.
What has gone wrong?
Shoddy Evidence: The premise that institutional deliveries would lead to decreased child mortality is based on correlations of institutional deliveries and child mortality. Based on such correlations, the United Nations decided that, “countries should use the proportion of births assisted by skilled attendants as a benchmark indicator” to monitor progress and aim for 80% of births assisted by skilled birth attendants by 2010. This is wrong on at least two fronts. First, as always, correlations are not causality.
In rich countries, any drug has to go through at least 6 years of testing prior to clearance, but when it comes to the policies that affect the bodies of poor women, can we not do better than relying only on associations in the data? Second, the marginal revolution transformed economics more than 100 years ago, but it seems to have had no sway on global health policy. You cannot decide what is optimal without understanding the marginal costs and benefits. (Simple example: if the cost of increasing institutional deliveries to 70 is $10, to 75% $12 and for every percent above $10 million, and additional benefits were the same, is 80% the magic number? And do we know anything at all about the shape of that marginal cost curve? No.)
Lack of respect for the decisions of the poor: Furthermore, government encouragement of institutional deliveries is based on the idea that poor people choose to deliver at home either out of ignorance or an inability to make the right decisions or due to cultural norms and the exercise of (male) power. But an alternate starting point is that people were not using institutions to begin with precisely because quality was low, and that increasing quality would also bring more people in. In fact, this is the most obvious explanation for the correlation between increasing institutional deliveries and lower child mortality. To base policy on the belief that we can make better decisions over the lives of those who are about to be born than their parents is a stand that minimally requires the onus of proof to be on those who claim such knowledge.
Of course, there are situations where women cannot leave home to give birth and where patriarchal power is played out over the birth of a child. But these are cases that have to be worked on piece by piece and not assumed to be the constraining factor in the general population. The likely reason why the community health center was packed and the primary health center was not was that our young obstetrician really cared and tried to do her best, relative to the doctor who just did not care.
Lack of contextual understanding: Especially, when it comes to disadvantaged populations who do not have a voice of their own, we must advocate on the behalf of evidence and an understanding of context. In India, the international drive to increase institutional deliveries was juxtaposed with a continuing obsession around family planning that has turned many public clinics into mass sterilization wards. This was a predictable outcome based on a cursory reading of the Indian government’s obsession with family planning.
Too often, we are seduced by the fallacy of immediacy (rigorous evidence will take too long) or the hubris of our own certainties (we know what we need to do). Let us also have the humility to accept the harm that has resulted and learn from these costly failures. Where the evidence is not clear or is not there, we should always take the time to set up the accompanying research—even when there is a global consensus that something needs to be tried due to extraordinary circumstance. And where we do not have the evidence, let’s be honest that we do not know.
Bertrand Russell once wrote: “There have been ages when everybody thought they knew everything, ages when nobody thought they knew anything, ages when clever people thought they knew much and stupid people thought they knew little, and ages when stupid people thought they knew much and clever people thought they knew little. The first sort of age is one of stability, the second of slow decay, the third of progress, the fourth of disaster.” (Russell: On Modern Uncertainty).
He believed that we are now in the fourth age; perhaps it is time to prove Russell wrong.