How and why do countries vary so much in their use of health services?


This page in:

I’ve been struck recently by how little we (or at least I) seem to know about variations in use of health services across the world, and what drives them. Do people in, say, India or Mali use doctors “a lot” or “a little”. Even harder: do they “overuse” or “underuse” doctors? At least we could say whether doctor utilization rates in these countries are low or high compared to the rate for the developing world as a whole. But typically we don’t actually make such comparisons – we don’t have the numbers at our fingertips. Or at least I don’t.

I’m also struck by how strongly people feel about the factors that shape people’s use of services and what the consequences are. There are some who argue that the health problems in the developing world stem from people not getting care, and that people don’t get care because of shortages of doctors and infrastructure. There are others who argue that doctors are in fact quite plentiful – in principle; the problem is that in practice doctors are often absent from their clinic and people don’t get care at the right moment. There are others who argue that doctors are plentiful even in practice and people do get care; the problem is that the quality of the care is shockingly bad. Who’s right?

WHS to the rescue – again

As in a recent post of mine on Let’s Talk Development, I thought the World Health Survey might shed some light on these issues. The WHS was fielded in the early 2000’s in 70 countries – spanning the World Bank’s lower-, middle- and high-income categories. The WHS enumerators asked a randomly-selected adult in each household about his or her use of inpatient care and outpatient care; in the numbers that follow I’ve focused on use in the last 12 months. As I said in the earlier blog post, the WHS does have some drawbacks: it covers some regions fairly fully, other much less fully; it’s 10 years old; and all we can tell is whether inpatient or outpatient care was received, not the number of contacts. But despite these problems, the WHS gets us quite a long way.

A lot of variation – but not necessarily what you’d expect

The maps below show the inpatient admission and outpatient visit rate – actually the fraction of people who had at least one admission or visit in the last 12 months. Green countries are above the developing-country average; red countries are below it.
For IP admissions, most of the OECD countries are above the developing-country average (6.98%). Brazil, Namibia and the European and central Asian countries are also above it. African and Asian countries are mostly below or close to the developing-country average.

The picture is different for outpatient visits. Several OECD countries are actually below the developing-country average (27.52%). And for the most part, the countries below the developing-country average are in Africa: many are considerably below it (Mali stands out dramatically); only a few are above it (Kenya and Zambia stand out). By contrast, several countries in Asia are above the developing-country average: India and Pakistan are dramatically above it, but China and Vietnam are also above it; a few Asian countries are below it – Laos and Myanmar are considerably below it, Malaysia and the Philippines less so.

Do variations in doctor numbers and infrastructure explain variations in utilization?

The maps below show data on doctors and hospital beds per 1,000 persons. I got the data from the World Development Indicators, and took the country averages for the first half of the 2000s. As before, green countries are above the developing-country average; red countries are below it. The countries above the developing-country averages are mostly those in the OECD and Europe and central Asia, though in the case of doctors per 1,000 some of them are also in Latin America and the Caribbean. Except for China, most of Asian countries fall below the developing country average.

Correlating the WHS utilization data with the WDI doctor and beds data shows that doctors and beds per 1,000 persons are positively associated with outpatient visit and inpatient admission rates. A lack of doctors and beds looks like it could indeed be part of the explanation for low utilization rates, though of course we haven’t established causality.

But a lack of doctors and hospital beds is only part of the story. Together they “explain” only 60% of the cross-country variation in inpatient admission rates, while doctors “explain” an even smaller 20% of the cross-country variation in outpatient visit rates.

Some countries – India and Pakistan are examples – are below the developing-country average on doctors per 1,000 persons, but above the developing-country average on the outpatient visit rate. Doctors and hospitals in these countries treat far more patients than one would expect given the number of doctors and hospital beds in these countries. In these countries, it doesn’t look like accessibility is the pressing issue; as research by my colleague Jishnu Das confirms, at least in India, poor quality is the bigger problem.

By contrast, much – but not all – of Africa is in the opposite camp: these countries have inpatient admission and outpatient visit rates that are below what would be expected on the basis of their doctor and beds per 1,000 figures. So it’s not just that these countries lack doctors and beds; it’s also that people are not getting the level of contacts you’d expect from the existing staff and infrastructure. Here it looks like absenteeism could well be part of the story; recent research from my colleague Markus Goldstein confirms it – pregnant women whose first clinic visit coincided with a nurse’s attendance were found to be 46 percent more likely to deliver their baby in a hospital.

Two take away messages

Message #1 is that countries differ considerably in their utilization rates. Much of Asia visits doctors more regularly than both the developing world and the entire world; India’s consultation rate is a third higher than the global average. Africa stands out as the continent where outpatient visits and inpatient admissions lag behind the rest of the world.

Message #2 is that these variations are partly explained by differences in doctors and hospital beds per capita, but only partly. The problem goes deeper than hiring more doctors and building more hospitals. Africa has lower outpatient visit rates than its doctors per 1,000 figures would suggest, while the opposite is true of India and Pakistan. In Africa, it looks like the binding constraint may well be absenteeism, while in S Asia it looks like the first-order problem is the poor quality of care that’s actually delivered.



Adam Wagstaff

Research Manager, Development Research Group, World Bank

Join the Conversation

Mohibul I Priyo
December 11, 2013

Your hypothesis about utilization rate in South Asia matches ("Doctors and hospitals in these countries treat far more patients than one would expect") with my experience in researching healthcare service delivery in Bangladesh (which seems to be similar in this regard to Pak and India).
I found that patients visiting state hospitals wait for 2-6 hours in line to get 1-3 minutes of attention from state doctors, who end up viewing 200 patients approximately in a 4 hour stretch.
While this system is cheap and provides basic care to large volume of rural poor, complicated cases are not not given the care needed.

Salman Khan
December 12, 2013

Hi Adam, Thanks for writing this great blog. Thoroughly enjoyed it!
Just one query. How excatly are you defining a "doctor" here? This is relevant because in India and Pakistan we have a lot of quacks and a lot of people do go to them! So in the WHS, what's the definition used? If the quacks are included, it could explain a significant part of the low quality of services.
Moreover, it is my impression that a major problem in countries such as India and Pakistan is that people don't go for "cheap prevention" but rather go for "expnesive cures". That could also explain the high number of visits to doctors and them serving a greater than expected set of patients.

Adam Wagstaff
December 18, 2013

Salman, thanks. The WHS asks about the last "health provider" visited, and then asks what type of provider, with one category being "traditional medicine practitioner". On quality of different types of providers in India, see Jishnu Das's article that I linked to in the blogpost—it shows the quality issue isn't as straightforward as people often think it is. A.  

Pieter Walhof
December 12, 2013

Dear Adam,
As always your analysis is very thorough. My question would be what the cause is behind the difference. Maybe we should include in the analysis the economic context and find out the differences in how transactions occur in healthcare?
Healthfinancing systems, or the lack thereof, the investments in healthcare, consumption indicators, medical expense inflation, government expenditure in relation to ODA, medical tourism expenditure can give a view on how a health market is performing on an economic basis.
Example: if investments are low or decreasing, it gives a view on how risk factors develop?
Hope this is a valuable amendment and look forward to your future research!
Pieter Walhof

Adam Wagstaff
December 18, 2013

Pieter, thanks. Emre below has some interesting ideas, but pls also see my reply to him on the limited scope of the post. A.    

Matthew Greenall
December 12, 2013

Might people in the countries with lower usage rates also be getting care from other sources? Community health workers, traditional med, patent medicine vendors, pharmacies are all important sources of care that I suspect are not fully captured in the WHS.

Adam Wagstaff
December 18, 2013

Matthew, thanks. Pls see reply to Salman above. A. 

Emre Ozaltin
December 12, 2013

Dear Adam,
A very interesting look at the links between utilization and HHR and health infrastructure.
In addition to availability of services, some thoughts on further layers readily available from WHS and other sources that could be added to the analysis of utilization:
Health status – while endogenous, the links between aggregate levels of health (self-rated, based on coverage indicators, or based on other aggregate indicators at country level) and utilization may yield interesting insights.
Met/unmet need – beyond levels of utilization, the WHS allows us to measure the proportion of those who needed care who sought care (or the converse). We have this overall and for a number of conditions. The unmet need may be more strongly correlated with service availability than utilization overall.
Levels of social spending - as much of the determinants of health lie outside the health sector, this variable may account for a substantial portion of the variance in utilization, for example the relatively low utilization in Europe. Elisabeth Bradley at Yale is doing interesting work on this in developed countries, with potential implications for the developing world.
Prevention – similar to the previous variable, indicators at country level on prevention activities at the population level, such as legislation related to smoking, drinking, diet and road safety as well as health promotion activities may account for some of the variance in utilization beyond service availability.
Inpatient/outpatient – the WHS allows us to disaggregate utilization by inpatient/outpatient (unfortunately we cannot do so by level of care). There may be interesting differences at this level of disaggregation that yield insights and patterns by country.

Adam Wagstaff
December 18, 2013

Thanks, Emre. The last part of the post was really little more than some exploratory correlations to see how much (or little!) of the cross-country variation in utilization is explained by availability. In that sense the title was a bit misleading, promising more than was actually delivered! Your thoughts on what the other explanatory factors might be — and where the data might come from — are great. Thanks for sharing. A.