The assessment of the prognosis of musculoskeletal conditions in older adults presenting to general practice: a research protocol

The primary objective of this study is to determine the prognostic value of a brief assessment in adults aged 50 years and over presenting to general practice with musculoskeletal conditions.

• Develop a brief, practical tool for assessing pain in older people presenting with musculoskeletal conditions in primary care

• Test its feasibility and practicality for use in the general practice consultation and its acceptability to general practitioners

• Determine its ability to predict subsequent outcome

• Provide primary care clinicians with a simple standard method to rapidly identify patients with different prognoses, alongside their usual clinical assessment and judgement

• Encourage primary care clinicians to attend to prognostically relevant features of pain and disability as well as to traditional biomedical characteristics of musculoskeletal pain in older people

• To provide the basis for future research into the usefulness of such pain classification for patient care and treatment selection

Prospective observational cohort study in general practice.

Older adults (aged 50 years and over) registered with six general practices presenting with an episode of musculoskeletal pain to the general practitioner between September 2006 and March 2007.

All recorded general practice consultations for musculoskeletal pain among adults aged 50 years and over in 6 general practices in Central Cheshire occurring between September 2006 and September 2007 (participants may only be sampled once).

Potentially eligible participants will be identified in the consultation by the activation of an electronic template when a selected musculoskeletal Read code is entered into the Egton Medical Information System (EMIS). On activation of the electronic template, general practitioners may use their discretion to apply the exclusion criteria as appropriate and exit the template. If not excluded, the electronic template will remind general practitioners of the study and prompt them to record their responses to seven brief questions. The consultation will continue (e.g. with advice, treatment) as per normal. The consultation record of that patient will be electronically "stamped". The template will not be activated if the patient has already had an eligible consultation within the study period, thereby ensuring that patients can only be sampled once. The electronic template is presented in Figure 1. The following questions are asked by the participating GP:

In addition to the six questions given in Table 1, the GP is also asked for their opinion on the likely prognosis for the patient, using the question below:

What do you think the outcome of this patient will be in 6 months time? (response items: completely recovered, much improved, improved, same, worse, much worse).

Weekly electronic searches by members of the Research Network Team will identify all eligible patients in the preceding week with a stamped record. Eligible patients' names and addresses will be downloaded into a secure mailing database. No other information will be accessed unless and until written informed consent to do so is obtained from the patient.

Throughout the entire mailing process, the Research Network team will perform weekly checks for patient deaths and departures, to ensure that patients do not get inappropriately contacted. Participants returning their questionnaire and consent sheet will be logged on the database and no further reminders will be sent.

Eligible patients will be sent a Study Pack from their general practitioner, containing a Patient Information Sheet outlining the study, inviting them to take part, and telling them what they have to do if they would like to take part. The name and contact telephone number of the Principal Investigator will be provided should potential participants have any questions about the study. The Study Pack will contain a Baseline Questionnaire with Consent Form and a stamped addressed envelope.

Non-responders to the mailed Study Pack will be sent a Reminder Postcard at two weeks and a Reminder Letter with repeat Baseline Questionnaire two weeks later. Non-respondents after all three baseline mailings will be assumed not to have consented to take part and will not be contacted again. The mailing database will record non-respondents and non-consenting responders (either by telephone or post) to ensure that patients are not invited to take part more than once during the study period. Participants who consent to be followed up by postal questionnaire but do not want their medical records accessed will receive mailings at three, six and 12 months but will not have their medical record accessed.

Participants who return their Baseline Questionnaires and provide written informed consent to further contact and to accessing their medical record will be sent Follow-up Questionnaires at 3, 6, and 12 months after their index consultation.

A flowchart of the process of recruitment is provided in Figure 2.

The other source of data collection in the study is self-complete postal questionnaires. This provides information on the descriptive characteristics of participants, outcome measurement, and more detailed prognostic information to supplement that collected from the brief prognostic assessment in the consultation. The conceptual domains, operational definitions, and empirical measures are provided below in Table 2.

Data will be entered into a database specifically designed for this study. Prior to data entry, this database will be tested using a set of dummy data. Data will be entered as the completed questionnaires are received by dedicated members of the administration team. Although they are experienced in data entry, specific training will be provided for this study. The index site of pain triggering the initial consultation will be entered into this database, so that subsequent mailings can refer to this index pain.

The lead investigator and the study statistician will determine coding prior to data entry. The database will provide coding options, to facilitate the entry of data. Some standard codes (e.g. missing data (-9), not applicable (-88)) are used by the Centre and will be utilised in this study.

A different member of the team will then check 1 in 10 random questionnaires as part of a quality assessment process. This information is kept by the research support co-ordinator and the study statistician.

Data will be stored in a password protected database. Only members of the research team will have access to the data. Requests for access to the data must be made in writing, along with an analysis plan, to the custodian of the data. Questionnaires and consent sheets are securely stored in separate locations, to protect the confidentiality of participants.

Using data from the 1991 National Survey of Morbidity in General Practice and more recent data from the North Staffordshire Primary Care Research Consortium's database of consultations, we estimate that the frequency of consultation for musculoskeletal conditions in those aged 50 years and over is 7 per 1000 registered patients per month (discounting repeat consultations within the same year). The approximate practice denominator population in this age-group for the 6 participating practices is 18,000. This means that the potential numbers of recruited patients over a 3 month period would be approximately 380. However we have to allow for a combined non-response and non-consent to medical record review of 20%. This will leave an estimated 300 participants completing 12-month follow up.

Data from the three sources (brief assessment tool, general practice medical records and self-complete questionnaire) will be analysed as follow:

I. Descriptive account of flow of participants: eligible, mailed, responded, consented, followed up

II. Descriptive account of completeness of data and comparison between brief prognostic assessment and self-complete versions of items

III. Descriptive characteristics of participants, including comparison between completers and non-completers.

IV. Description of frequency of recovery and non-recovery at 3, 6, and 12 months among completers.

V. Univariable comparison between GP prognostic judgement at baseline and recovery at 6 month follow up (Cox regression).

VI. Univariable comparison between patient prognostic judgement and recovery at 6 months (Cox regression).

VII. Univariable relation between brief prognostic assessment variables and recovery at 3, 6 and 12 months (Cox regression).

VIII. Multivariable relation between brief prognostic assessment variables and recovery at 6 and 12 months (Cox regression).

IX. Comparison of prognostic accuracy of GP judgement, patient judgement, and brief prognostic assessment for predicting recovery at 6 months

X. The marginal informativeness of brief prognostic assessment variables to GP prognostic judgement for predicting recovery at 6 months (multivariable Cox regression).