Measuring the clinical course of low back pain : using course and indications for care to identify subgroups

Abstract

Background: Non-specific low back pain (LBP) is a very common and costly condition and it is recurrent in a large proportion of cases. However, little is known about the detailed course over time and whether the course varies between individuals or groups of patients. Presumably, several subgroups exist, each with a different course and treatment needs. Exploring the detailed clinical course of LBP is possible with a new method of frequent data collection, short text messages sent and received with mobile phones.

Aims: The overall aim of this thesis was to explore a new method of collecting frequent data by text messaging used to define the clinical course of non-specific LBP in subjects seeking treatment in the primary care sector. The specific objectives were 1) to evaluate this text messaging method, 2) to illustrate various ways of analysing this type of repeated data, 3) to identify clinically relevant clusters on the basis of the clinical course, 4) to investigate the association of common baseline variables with outcomes based on frequently measured data, and 5) to identify clinicians’ opinions of the indications for secondary and tertiary prevention of recurrent and persistent LBP.

Summary of methods: In two prospective longitudinal studies, chiropractic LBP subjects were followed for six months/ 18 weeks with weekly text messages. The use of text messages as a data collection tool was scrutinised in terms of response rate, user friendliness and compliance. Several methods of analysing repeated data were evaluated and illustrated in a model data set with LBP subjects from these studies. The clinical course of LBP over six months was used to subgroup subjects using hierarchical cluster analysis. The identified subgroups were described in terms of clinical course and demographic characteristics. The association of baseline variables with outcomes based on weekly text message data was explored. Preliminary focus groups and a questionnaire survey among Swedish chiropractors determined indications for recommending secondary and tertiary care to patients with LBP.

Summary of results: Using text messages and mobile phones to collect frequent data resulted in a high response rate, 82.5 %. Good user-friendliness is assumed as the drop-outs did not mention the method itself as the reason for discontinuing their participation, and these individuals were not a homogenous group regarding age, gender and LBP characteristics. The method showed high compliance rates; over 70 % of the respondents answered more than 80 % of their text message questions. The appropriate method of analysis will depend on the research question and characteristics of the outcome data. Several methods of analysing repeated data showed that all methods were robust concerning the association of a selected baseline variable with the outcome. Individual courses of LBP over six months showed great variety. The explorative cluster analysis grouped subjects into four clinically relevant units based on their LBP development over time. In predicting future LBP, all the examined variables interacted with time. Previous duration of the LBP condition predicted LBP at all the investigated time points after the first week. Focus group discussions and the subsequent questionnaire survey identified that the indication for secondary preventive treatment was the presence of previous episodes and that the indication for tertiary prevention was that the treatment was deemed effective.

Conclusions: Text messages can be used to gather frequent data prospectively on large populations. This method has advantages over traditional data collection methods and can be used when repeated information is warranted or when monitoring populations over time. Clinically meaningful clusters could be identified on the basis of course, but these subgroups need further exploration and replication in different populations with more clinical variables added. Similarly, as the predictive ability of some usual clinical baseline factors varies with time, predictors of future LBP needs exploration in detail. Knowledge about subgroups and indications for prevention possible strategies could be used to study the effect of such strategies.