Reappraisal of the Inverse Care Law – Methodology


The purpose of this study was to investigate whether the Inverse Care Law is still relevant to discussions of healthcare services today. Having found substantial evidence that the inverse relationship continues, despite resources being more concentrated in deprived areas, it remains to be explored whether this enduring problem is due to market forces as Hart originally suggested and is thereby likely to be affected by the level of deprivation of the area in which the health services are placed, or whether, as the literature suggests, the source of the Inverse Care Law actually lies in the GP-patient relationship and is therefore more likely to be affected by an individual’s socioeconomic class.


For the National Survey of Local Health Services 2006 eighty people registered with a GP and aged 16 or over were selected from each of the 303 PCTs (now merged to 152); questionnaires were sent to a total of 24,240 people. There was a 43% response rate overall therefore the sample population of the data used totals at around 10,000.

This particular survey was chosen for two reasons: firstly the questions were sufficiently relevant to explore respondents’ experiences of health services, especially key areas mentioned in previous studies, such as access and time waited for an appointment, consultation length and referral to specialist services. Secondly this survey was one of the few surveys to include a variable that could be used as a proxy measure for socioeconomic status; in this case ‘age left full-time education’; though socioeconomic status encompasses many aspects, education level is a good indicator of this (Taylor and Field, 2003:45).


The instruments used for this study consist of SPSS software, the National Survey of Local Health Services 2006 and the Indices of Deprivation 2007 PCT Summaries.


Structure of Study

A suitable survey was searched for that enabled healthcare experience to be measured in relation to both socioeconomic class and PCT area. Once an appropriate survey was found the results were statistically analysed using six health care indicators in relation to age and socioeconomic status, in the form of contingency tables and chi square statistics. The Indices of Deprivation 2007 PCT Summaries (see Appendix 2) was then consulted to select a sample of PCTs ranked both highest and lowest in terms of deprivation. Those included in the dataset were then selected and two groups were constructed within the sample, one of high deprivation PCTs and one of low deprivation PCTs. Clustered bar charts were then produced to compare the two groups against the six health care indicators.

Methods Used

The methods used for this study were solely quantitative; whilst a number of contacts were made with health professionals and experts, including Hart himself (see Appendix 3) interviews were not possible due to the restrictions of the Research Ethics Committee over access to the NHS employees and patients.

For these reasons (as well as lack of resources, limited time and lack of a socioeconomically and geographically diverse sampling frame), primary data collection was not possible. However in using a national dataset a much larger sample was obtained and therefore was more representative.

In using the National Survey of Local Health Services six indicators of the quality of health services were selected in relation to evidence in the literature review. The associated questions are as follows:

  • Indicator 1: ‘Last time made an appointment, how long waited?’
  • Indicator 2: ‘Were you given enough time to discuss your medical problem with the doctor?’
  • Indicator 3: ‘Did the doctor treat you with respect and dignity?’
  • Indicator 4: ‘In the last 12 months have you been referred to a specialist?’
  • Indicator 5: ‘Was the main reason you went to your GP dealt with to your satisfaction?’
  • Indicator 6: ‘In the last 12 months have you been put off going to your GP because of inconvenient opening times?’

As all questions are closed it was possible to present the data numerically.

Statistical Analysis

The data for the selected questions was then analysed in relation to ‘age left education’ and PCT high or low deprivation group. Those still in full-time education were filtered out of the sample as they did not reveal any level of socioeconomic status and so as not to distort the results. As age was recognised as a possible confounding variable the answers were also analysed in relation to age group to detect whether or not age reinforced the relationship between socioeconomic status, or obscured it.

The data was analysed using crosstabulation tables in relation to education level and age group, as well as chi square statistic; the level of probability considered significant was 0.05. Clustered bar charts were used to compare the results for high and low deprivation PCT areas.


As a lot of the questions in the survey relied on the subjective evaluations of the respondent, which may in turn be affected by their socioeconomic status in the first place, observations would have presented a more ecologically valid picture. However the Ethics Research Committee prevents any such research being undertaken within the NHS without much more time and persistent requests. Also there are issues of invading patients’ privacy.

A limitation of using secondary data is that there was no control over how variables were grouped, for example it would have been interesting to see how those under the age of 25 rated services, but the ages were already grouped 16-35. Also only one question was included that could indicate socioeconomic class (age left education), but there was a very high percentage (59.8%) missing from this question and it included those still in full-time education, without any reference to age and so socioeconomic class could not always be inferred.

Not all PCTs responded to the questionnaire and obviously not all patients at the included PCTs responded and it is likely that there is something different about respondents missing from the sample that may change the picture currently presented. As all questions were closed there may be some answers chosen arbitrarily if none of the options expressed the true response of the respondent.


The use of secondary data side-stepped many ethical issues: Informed consent had already been given by those who answered the survey and the anonymity of respondents was maintained. Similarly de-briefing and the right to withdraw did not have to be considered. Respondents retained their right to confidentiality and the survey was already published online and therefore classified as public knowledge.