I compiled these algorithms for The National Health Measurement Study, a nationally representative survey of the United States which co-administered the following instruments. More information about the survey and associated projects can be found at their website. Most of the algorithms here are from articles published in peer-reviewed journals. These articles are documented within the code. Some of the algorithms were provided by the instrument researchers and provided here with their permission, which is also documented within the code. I have been able to validate my code for some algorithms against various data sets. These validations have been listed on this page. However, It would be prudent to check my code against the original articles before using it as well as confirming that there have not been more recent updates to the algorithms. If you find an error, I would appreciate hearing about it. Please note that I have my code under a Creative Commons License. The specific terms of this copyright are available by clicking the link at the bottom of this page.

EuroQol Index (EQ-5D) The EQ-5D is a free instrument with a 5-item descriptive system and a visual analog scale. This instrument is maintained and developed by the EuroQol Group. The 5 item responses can be weighted to give a health related quality of life score. There are different weights available. For this study, we used the UK time trade off (TTO) weights and the US TTO weights. The UK TTO weights come from Dolan 1997 and SPSS code provided by the EuroQol group. I have checked this code against the 2000 Medical Expenditures Panel Survey, which included item responses and a calculated EQ5D-UK score. SAS 8.02 Code STATA 8.2 Code The US TTO weights come from Shaw et al 2005. This code was generously contributed by Jim Shaw and is shared with permission. SAS 8.02 Code STATA 8.2 Code Note: there are also predicted EQ-5D scores from the SF-12 in the Predicted Scores section below.

HALex The HALex is a post-hoc measure constructed from items in the National Health Interview Survey. This measure is interviewer administered with many skip patterns; though there are 14 questions in the measure, the maximum an individual will answer is 7, with most respondents answering just 3 or 4 questions. This code currently calcuates a HALex score if the responent answered the first general health question. If your dataset has many "don't know" or "refused" responses, this code may not be appropriate to use. SAS 8.02 Code STATA 8.2 Code

Health Utilities Index (HUI) Mark 2 and 3 The HUI instrument is proprietarily held by Health Utilities Inc. While the instrument is proprietary, the weights for the various dimensions are open for use by anyone. I have included code here that maps from the domain scores to the final summary scores. I have also previously written code that maps from question responses to domain scores if you would like to discuss them. Please license their questionnaires before using them. SAS 8.02 Code (Interviewer version) STATA 8.2 Code (Interviewer version) SAS 8.02 Code (Self-administered version) STATA 8.2 Code (Self-administered version) Note: There are also predicted HUI scores from the SF-12 and SF-36 in the Predicted Scores section below.

SF-6D The SF-6D is a recently developed scoring algorithm from a subset of questions in the SF-36 or SF-12 (see below). Eleven questions are used from the SF-36 while 7 questions are used from the SF-12. These questions were used to construct health scenarios that were evaluated using the standard gamble technique in a representative sample of the UK population. The algorithm assigns a utility based score to each health state. This code uses the consistent models in Table 4 from the 2004 Medical Care article for both the SF-12 and SF-36 based SF-6D (previously available code on this website was not from this article). A different model for the SF-36 based SF-6D had been published in Health Economics (2002), but was replaced with the model in Medical Care. If a researcher only wanted to collect SF-6D scores and not the SF-12 or SF-36 profiles, I believe they could use only the SF-6D items because the valuation was built on these questions alone and the order effects that might be present in the full surveys are probably small. I've included examples of what these "SF-6D Questionnaire"s might look like with each scoring algorithm.. SF-12 version 1 Example Questionnaire SAS 8.02 Code Stata 8.2 Code SF-12 version 2 Example Questionnaire SAS 8.02 Code Stata 8.2 Code SF-36 version 1 Example Questionnaire SAS 8.02 Code Stata 8.2 Code SF-36 version 2 Example Questionnaire SAS 8.02 Code Stata 8.2 Code

SF-12 The SF-12 is a subset of questions from the SF-36 (see below). Information about version 1 can be found here, here, and here. With the limited number of responses to certain questions in version 1, many people suggest that only the mental and physical composite scores be calculated, not the 8 subscales. The SF-12 version 1 was included in the 2000 to 2002 Medical Expenditures Panel Survey's Self Administered Questionnaire. An example of the SF-12 version 1 is questions 17 to 28 in the 2001 version. Because it is somewhat difficult to find a copy of this questionnaire, I have reproduced it here. SAS 8.02 Code STATA 8.2 Code The SF-12 version 2 uses essentially the same questions as version 1 with a few wording and response category modifications. Because of the increased number of response categories, it is considered appropriate to report both MCS/PCS scores and the 8 subscales. This version of the SF-12 is proprietary and held by QualityMetric Inc. Users must pay licensing fees to administer and score these questionnaires. Ron Hays provides SAS scoring code on his website. I have also implemented the scoring and would be happy to converse with others who are trying to implement it themselves. The Veterans-12 (VR-12) is another version of this questionnaire which is publicly available. Information about the VR-12 can be found here. Note: There are a number of predicted scores from the SF-12 version 1 below.

SF-36 The RAND-36 was originally developed by RAND for the Medical Outcomes Study. The RAND-36 and SF-36 version 1 are the same instrument though one question is scored differently between the two versions. Both the RAND-36 and SF-36 version 1 are freely available. A copy of this questionnaire can be found here at the RAND website, as well as scoring information and terms of use. The code below uses the SF-36 version 1 scoring. Ron Hays, one of the developers, also provides SAS scoring code on his website. SAS 8.02 Code STATA 8.2 Code The SF-36 version 2 uses essentially the same questions as version 1 with a few wording modifications, but expands the response categories from 2 possible responses to 5 possible responses for 7 questions and reduces the response categories from 6 possible responses to 5 possible responses for 9 other questions. This version of the SF-36 is proprietary and held by QualityMetric Inc. Users must pay licensing fees to administer and score these questionnaires. Ron Hays provides SAS scoring code on his website. I have also implemented the scoring and would be happy to converse with others who are trying to implement it themselves. Note: There are a number of predicted scores from the SF-36 version 1 below.

Quality of Well-Being Scale (QWB) - Self Administered v1.04 The QWB-SA is a free instrument with 77 questions that fall into 4 domains. The QWB was originally developed as an interviewer administered questionnaire with a 6 day recall. The self-administered version is newer and focuses on the past 3 days. I have provided a copy of the instrument here and a copy of the user manual is available at the HOAP website. SAS 8.02 Code STATA 8.2 Code Note: there are also predicted QWB scores from the SF-36 in the Predicted Scores section below.

Predicted Scores There are a large number of predicted scores from the SF-12 and SF-36. These predictions go from the SF family to utility based scales so that SF values can be converted and used in cost-effectiveness analyses. In general, it is preferable to use the recently developed SF-6D scoring algorithm (see above) over the predicted scored documented in this section. From the SF-12 version 1, there are predicted scores for the EQ-5D (with UK weights), HUI3, and VAS. All algorithms use the MCS and PCS composite scores, not individual items. Two of the six algorithms require age in years and sex. These algorithms are taken directly from the published articles, often with clarifications from the authors via personal communication. The articles and communications are documented within the code: SAS 8.02 Code STATA 8.2 Code From the SF-36 version 1, there are predicted scores from the QWB and HUI2. These algorithms use the 8 subscales of the SF-36, not the MCS and PCS composite scores. These algorithms are taken directly from the published articles, which are documented in the code: SAS 8.02 Code STATA 8.2 Code