Article Text

PDF

Erroneous, blurred, and mistaken—comments on the care need index
  1. Bo Burström1,
  2. Olle Lundberg2
  1. 1Karolinska Institutet/Stockholm County Council, Sweden
  2. 2Centre for Health Equity Studies, Stockholm University/Karolinska Institutet, Sweden
  1. Correspondence to:
 Dr B Burström
 Department of Public Health Sciences, Division of Social Medicine, Karolinska Institutet, Stockholm SE-171 76, Sweden
  1. Kristina Sundquist1,
  2. Marianne Malmström2,
  3. Sven-Erik Johansson2,
  4. Jan Sundquist2
  1. 1Flemingsberg Primary Health Care Centre, Sweden
  2. 2Karolinska Institutet, Family Medicine, Stockholm, Sweden

    Statistics from Altmetric.com

    Sundquist et al present a care need index for allocation of primary health care resources.1 Unfortunately, their paper rests on an erroneous description of the allocation model presently used in Stockholm, a blurred conception of need, and a mistake in the handling of data.

    The model used by Stockholm County Council to distribute funds between areas to purchase health care consists of four different components: (1) hospital based care,2 (2) private specialist care, (3) primary health care, and (4) pharmaceutical drugs. The primary health care model gives extra weight to neighbourhoods with high proportions of low income earners, immigrants, and single persons; and according to the proportion under 16 and over 64 years as they use primary health care more.3 This approach is as likely to capture health care needs in the population as the care need index (CNI) model, and it is not based on prior health care utilisation as suggested by Sundquist et al.

    In the CNI model “need” is defined on the basis of a set of pre-defined indicators that general practitioners have weighted according to their impact on GP work load. Models of health care utilisation usually differentiate between need and demand,4–6 as the probabilities to show up in the GP’s waiting room differ between persons and social groups, given the same need. GP’s experienced workload, however, is only affected by the patients in the waiting room; thus the theoretical basis for the CNI is demand rather than need.

    The empirical analyses are based on the annual surveys of living conditions. In these surveys the number of response alternatives to the self rated health question was changed from three to five in 1996, but the authors seem to treat the data as if there were three response alternatives throughout the period. As a consequence those with “good” health have been counted as ill in a third of the sample. This will cause the illness prevalence for 1996–97 to be overestimated and introduces a bias in the relation between health and other variables.

    The main practical consequence of applying the CNI rather than the existing model would be to “take from the poor to give to the poor”. Although the SS area is more deprived than the SW area according to the CNI a re-allocation from the first to the second is suggested. There must be more useful tools for allocation of primary care resources.7

    References

    Authors’ reply

    Burström and Lundberg claim that our article1 rests on (1) an erroneous description of the allocation model presently used in Stockholm, (2) a blurred conception of need, and (3) a mistake in the handling of the data.

    We apologise for the somewhat erroneous description of the present Stockholm model, although we believe that the allocation model presently used in Stockholm has several weaknesses. Burström and Lundberg declare that the present Stockholm model gives extra weight to neighbourhoods with high proportions of low income earners and immigrants. However, they define low income earners as men (women are not included) in the three lowest income quartiles and have not justified the reason for this broad definition of low income earners. In addition, immigrants are defined as foreign born people from all other countries in the world in contrast with Swedish born people. However, many immigrants in Sweden were born in western countries and have a similar health status to Swedish born people. Although we agree that the present model is not based on prior health care utilisation it is based on morbidity, defined as proportions of people with long term sick leave >30 days, which we assume have been taken from prior healthcare registers.

    They also claim that the conception of need in care need index (CNI) is blurred. We do not agree with that statement. CNI as well as UPA score include need based items in their modelling of the allocation of healthcare resources. These instruments for allocating resources to primary health care have defined “need” according to the higher need for health care among certain groups in the society. CNI includes weighted neighbourhood proportions of a total seven different demographic and socioeconomic items, such as people with low educational status, foreign born people from non-western countries, and single parents. Our article also shows a strong relation between CNI and self rated health, which is a good proxy for health care need in the population. Previous studies of CNI (13 original articles and two theses) have demonstrated a significant relation between CNI and different health outcomes, all relevant for primary health care.1–13 The documentation of the present Stockholm model is not that substantial.

    In addition, in their critique, statements about the GPs’ experienced workload and the GPs’ waiting room are included even though none of them are working as GPs. In contrast, three of the authors of our study are working as specialists in family medicine.

    We do not understand what underlies their statement that we were mistaken in the dichotomisation of the outcome variable. We have indeed noticed that the number of response alternatives to the self rated health question was changed from three to five in 1996 and have accounted for this in our study. The dichotomisation was performed as follows: Before 1996: Those who answered that their general health was bad or something in between were considered as having poor self rated health. Those who answered that their general health was good were considered as having a good health status. After 1996: Those who answered that their general health was very bad, bad, or fair were considered as having a poor self rated health. Those who answered that their general health was good or very good were considered as having good self rated health. If the response alternatives had been dichotomised as they claim, the associations would have been much weaker or even disappeared.

    Finally, Burström and Lundberg have referred to an article that was not published when we submitted our article.14 We agree that there are many other needs based capitation formulas. However, one of the advantages of CNI (or the Swedish UPA score) is the extensive documentation of CNI and different health aspects, such as utilisation of psychiatric hospital care, sales of tranquillisers and analgesics,5,7,8 unhealthy lifestyle factors that reflect an increased need for preventive efforts within primary health care10 and incidence of coronary heart disease.13 In addition, every county in Sweden is free to choose an appropriate tool for the allocation of primary healthcare resources. In accordance with our findings we conclude that CNI constitutes one such appropriate tool, based on the health care need in the population.

    References

    View Abstract

    Request permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.