Evidence Submission (Quantitative): Do the public really want more choice in the NHS? Using data from the British Social Attitudes Survey 2007 to verify Government claims.

Background

In April 2011, the UK Government released a leaflet called “Working Together for a Stronger NHS” 1.  The purpose of the leaflet is to explain to the public the health service reforms proposed in the White Paper “Equity and excellence: Liberating the NHS”2.  Shortly after its release, epidemiologist and Guardian columnist Dr Ben Goldacre wrote an article3 stating that the leaflet used statistics that were old or misleading.  One example is presented on page 11 which stated that “95 per cent of people want more choice over their healthcare”.  The leaflet references the 2007 British Social Attitudes Survey4 as the source of this claim.  The data for this survey is downloadable to people with a University account5.  Provided that users consent to an End User Licence agreement, the raw data can be downloaded and used to verify the statistics given in the Government leaflet.

From the raw dataset, there are two questions on choice with options of “a great deal”, “quite a lot”, “a little”, “none at all” and “don’t know”. (Bold text from original document):

Question 583: How much choice do you think NHS patients should have about the kinds of treatment they receive?

Question 584: And in your area, how much choice do you think they actually have (about the kind of treatment they receive?)

As Dr Goldacre points out, neither question asks people if they want “more choice” in the NHS.  At the same time, one of our team wrote a blog post6 which used a cross tabulation of the responses to the two questions above to attempt to reproduce the Government’s figure of 95%.  From this, it appeared that the figure must have been taken from the sum of the number of people who reported that they felt they “should” have “a little”, “quite a lot” or a “great deal” of choice about the treatment they received (although this actually totals 96%).  In this blog post, it is also noted that if we compare those who feel they should have more choice than they actually have, we might generate a figure for how many people want more choice.

Methods & Results

Building on this idea, we decided to investigate the data further.  We found there are further similar questions about choice, including: how much “choice” patients should have and actually have around treatment (Questions 581 & 582); how much “say” patients should and actually have over the hospital they attend (Questions 399 & 400); and how much “say” patients should have over the treatment they should and actually receive (Questions 403 & 404).

We each imported the data into separate statistical software packages and analysed the data by producing simple descriptive statistics for each question, and producing cross tabulations for each of the should and actually have questions.

The possible source of the 95% figure is from Question 399 (“How much say should NHS patients have over which hospital they go to?”): the total number of people answering “a great deal”, “quite a lot” and “a little” totals 2936 people out of 3078 who were asked the question i.e. 95.4% (Table 1; SPSS v17.0 export).  However, this still does not accurately reflect that 95% of people want more choice.

Table 1: The amount of say patients feel they should have over the hospital they attend

To attempt to obtain a figure for the number of people wanting more choice from this dataset, we chose to use the cross-tabulations to establish:

– Those people who felt they “should have” a given level of choice, and who believed they “actually have” the same level of choice.  We deemed that these people felt they have the “right amount of choice” (marked in yellow, Table 2; SPSS v17.0 export).

– Those people who felt they “should have” a greater level of choice than they “actually have”.  We labelled these people as having “too little choice” (i.e. these people would want more choice than they have) (marked in red, Table 2).

– Those people who felt they “should have” a lower level of choice than they “actually have”.  We labelled these people as having “too much choice” (marked in green, Table 2).

By totalling all of the red boxes, we found that 1246 people out of 2022 who were asked this question (61.6%) felt they actually had less choice than they should have.  Comparison with results from the other questions that we cross-tabulated generated a range from 61.2 to 72.0% depending on the specific question asked (see our Microsoft Excel spreadsheet).

This work has raised interesting questions for possible future secondary analysis of the dataset.  For example, the survey also asks people about their current state of health, as well as demographic information about age and area of residence.  Therefore, it might be possible to see if there are regional differences in attitudes to choice, or whether older people are more satisfied with the “choice” within the NHS than younger people, or if sicker people feel that they actually have a lot of choice compared to what they feel they should have.

However, we do not feel that this survey accurately measures how many people want more choice, as that was not the specific question asked.  For this to be accurately measured, a specific question should be asked within the survey.  This is echoed by the organisation that carries out the British Social Attitudes surveys, the National Centre for Social Research.  They state that: “There is widespread public support for the idea that people should be able to exercise choice when using public services…this does not necessarily mean the public want public services to be run by a diverse range of providers”⁴.  It is important to bear this in mind when reading the Government leaflet, but the restrictions on accessing the raw data limit the possibilities for members of the public to corroborate the facts presented to them.

Table 2: Comparing how much choice people think they should have about kinds of treatment with the amount they feel they actually have

Summary

It is important for the public to feel they can trust the data presented by the Government.  This work shows that secondary data analysis can enable statistics to be checked independently, provided access is available.  Members of the public without access may need to rely on those that do to maintain a critical watch on Government material so that they can hold them to account for their claims.

Authors:

Siobhan Farmer, Specialty Registrar in Public Health, North West School of Public Health

Anne-Marie Cunningham, Clinical Lecturer, Cardiff University

Mark Hawker, Teaching Development Fellow, University of Leeds

References and additional data

Additional BSA07 Choice Analysis