Researchers from Indian Institute of Management, Ahmedabad developed an index of Indian states based on their relative health indicator performances of their populace
A nation’s economic performance depends a lot on the heath of its populace. Not only does it improve efficiency but it is also an indicator of an all-round performance of the nation. Policy makers cannot ignore its importance in economic development as well as happiness and human development of the nation. Performance of health systems has been a major concern of policy makers in India for many years. Health has been made one of the most important dimensions of the millennium development goals set for a country. Four of the 18 development goals adopted by the General assembly of the United Nations (UN) in September 2000 relate to health and well-being of the population of the country. The focus of the Indian government is to develop a holistic care system that is universally accessible, affordable and effective’ is an indication of a national focus aligned to the UN goals.
This article is based on a study that explores the possibility of looking at the Indian states based on their relative performance on health indicators of their populace. The study involves collecting and purifying the data from reliable sources and developing an index, captures the variances across states and also gives direction for developing policies, strategies and action plans for each of the states. For a diverse country like India, it is important that the study provides a path for each of the state due to wide variation in terms of economic, geographic, social, cultural and political canvas.
The article discusses a unique outcome: Input matrix to develop insights on the relative status of states. It also indicates as to which states are performing better with lesser resources as well as those not able to utilise the resources to the optimum level.
The approach
The study started with collecting the list of variables used by WHO to assess the health performance. These included Infant Mortality Rate (IMR), Under 5 Mortality Rate (U5MR), Neo Natal Mortality Rate (NNMR), Maternal Mortality Rate (MMR), deaths due to HIV, deaths due to TB and deaths due to malaria. It was also found that WHO also uses a list of 47 diseases under the classification of communicable and non-communicable. Taking cues from some other studies which used sanitation and vaccination as parameters, a list of diseases was prepared. These indicators formed the outcome variables for health performance that were used to understand relative position of states.
Health is a state subject and also the states differed in their expanse and population characteristic. Each of the states also develops separate plans for themselves based on their priorities and any national agenda in line with the millennium goals. And each state tends to use a very different level of inputs to achieve its outputs. In a resource constrained nation, and consistent with sustainable development goals, any health index needs to also capture the role of inputs along with that of outputs. A list of input variables was, therefore, also prepared which consisted of infrastructure, manpower and economic factors. Based on the discussion within the team members, another dimension of utilisation of the service was added.
The study collected data from several sources like library, reports and databases as available from websites of government departments, development organisations and research and consulting companies. The effort was to collect data on as many items as possible for each state for the last five years (2008-2013). It was a critical step for choosing the items for further analysis.
Three factors were considered in choosing the sources and extent of data: (a) the data is available for each state, (b) it is available for five years and (c) the source is credible and acceptable to the stakeholders. Under these three conditions, some of the items were dropped. Diseases which had low incidences (less than 1000 per lakh population) were removed. The final list of variables used in the study is given in Table – 1. A similar method was used to select the states. In the final analysis, 21 states were chosen.
Data was collected for five years. A one year lag was considered for the outcome data. Thus the data for input items was collected for the year 2008 – 12, whereas, data for outcome items were collected for 2009 – 13.
Analysis
Two methods were applied to arrive at the scores for each of the states. In the first case, a regression analysis was used. When regressed on input variables, the results did not indicate any clear direction. The correlation matrix indicated that only a few of the input variables were correlated to the outcome. There were also incidences of multi-collinearity. We think that economic variables may also create a nested situation since the inputs created by the state would be a function of the budget allocated for health. Hence, this method was not pursued further.
In the second method weighted sum approach was used using factor analysis. Factor analysis helps to reduce the number of observed variables into smaller number of variables which account for the most of the variance in the observed variables. Analysis was carried out with all variables put together. The findings clearly hinted at the problem as was faced during the regression method. That is the relative position of states on notional index that combined both input and output indicators which was problematic from a diagnostic and a policy perspective.
So, finally it was decided to carry out factor analysis separately for input and outcome variables.
The analysis was carried out in two stages. In the first stage, weights of each of the items were determined based on the factor loading of each item. Using the normalised scores and these weights, a summated weighted score was arrived for each of the input factors. A second level factor analysis was carried out using the weighted scores for each factor for that dimensions. The factor loading so arrived was used to arrive at a summated score for the input dimension.
Findings Outcome
The normalised values and weights arrived from factor analysis of items of Child and Maternity Mortality rate (IMR, MMR, NMR, U5MR) a weighted summated score for the first parameter was arrived. Similarly for the values and the weights of the items of deaths due to other diseases (death due to HIV, TB, Pneumonia, and Acute Diahorrial Disease) a summated to score was obtained for the second parameter of outcome.
State ranking for outcome
Later, factor analysis was conducted on these two factors to attain the weights (factor loadings). These weights were used to arrive at a weighted score for the states for each of the five years. Based on the weighted score for the year 2013, the final scores weights were arranged in ascending order to arrive at the relative ranks of all the states as given in Table – 2.
Input
Similar to the above analysis, the normalised values of the items under parameters infrastructure, manpower and utilisation were summated to attain a score of the input dimension. Further, factor analysis was conducted on these three factors to attain the weights (factor loadings). These weights were used to arrive at a weighted scores and based on the weighted score for the year 2013, the final scores weights were arranged in ascending order to arrive at relative ranks of all the states as given in Table – 3.
Outcome: Input map
The study found that that there is a low degree of association between the input and outcome. Hence, the two ranks could not be compared or combined. It was decided to consider them separately. Using them as the two axes, an outcome – Input matrix was created (Figure – 1). Median of the dimensions was used to create the high and low quadrants. Since all ranking systems across the world use outcome measure for arriving at the scores, it was given higher importance as compared to the input measures.
The input: outcome matrix divided the states in four quadrants. It indicated that while resources are important, implementation of the initiatives to achieve the desired outcome is equally important. It also brings out that states and their policies need to be developed as suitable for the each of the states. It is expected that such segmentation would help in developing more focussed policies as well as fine tune health policies for states and also at a level of cluster of states. The states were divided into four classes:
High performers
Maharashtra, Tamil Nadu and Kerala are high performing states. These states are performing well with respect to the inputs made available or created in the state. Kerala scores highest in outcome but falls to a lower rank due to resources.
Middling performers
Jammu and Kashmir, Delhi, Andhra Pradesh, West Bengal, Himachal Pradesh, Gujarat, Punjab, Haryana and Karnataka and lie in the low input and high outcome quadrant. Their performance falls in the middle range.
Strugglers
Jharkhand, Bihar, Chhattisgarh, Orissa, Madhya Pradesh, Rajasthan, Uttar Pradesh, Uttaranchal and Assam are making efforts to come on the higher outcome performance. Of these states, Uttaranchal, Assam and Orissa are the poorest in terms of the resources available to them as well as lag behind in terms of come.The states were then ranked accordingly. The final ranking of the states based on the quadrant mapping can be seen in The five year performance based on the outcome: input matrix and the movement of states in terms of their performance on health relative to one another can be seen in Figure – 2.
The way forward
The Outcome: Input matrix shows us the relative status of states and also which states are performing well and which are not. There are some states which are not performing well because the resources available to them is limited. Very few states are performing well. This study can be used by the government to analyse the current position of the states and identify focussed policies to be worked upon. It provides a snapshot of where policy attention and implementation details need to be looked at with more focus.
The current study has an enormous scope for further research in analysing the performance of the states. Since the data was not available for all the states; more updated and adequate data can be incorporated in further research. Some new variables can be included in the outcome and input parameters depending upon the availability of the data from the secondary as well as primary sources. This report is a first step. With the existing data, we are able to make limited comments on the drivers of changes – which can provide pointers on where action is required. As we take up the next step, we will be able to bring our directions at more granular levels. The goal would be to have an annual index that is able to show how states are moving relative to one another.
- Advertisement -
Comments are closed.