Information in a Dutch GP registration system about high risks on long-lasting sickness absence: A cross sectional exploration in their registration system

Introduction

Long-lasting sickness absence (LLSA) is an important and growing problem that has consequences for individuals and society [1]. It leads to ‘reduced probability of eventual return to work and subsequent economic and social deprivation’ and is in itself unhealthy [2]. For society it leads to productivity loss and high costs for social benefits. So prevention in an early phase of LLSA is important. Doctors are involved in several ‘domains’: diagnosis and treatment, prevention, rehabilitation and return to work, and in certifying [3]. In the domain of treatment general practitioners (GPs) may play an important role in prevention because they often see the patient/employee in an early stage of sickness. They can recognize high risks of LLSA early and take these into account in their integrated care. But before the GP can perform these tasks systematically, they have to ask their patients actively about their work situation and high risks, and record the information. The question is to what extent this information is recorded in GP information systems.

Methods

In 2000, we performed a literature search in Medline using the search terms: high risk of sickness absence, GP, long-term sickness absence, indicators, determinants, prognostic factors. In addition we searched the Dutch literature by hand using the same search terms. From the literature thus retrieved [1], [4–7], we concluded that the following factors are likely to predict a LLSA exceeding 3 months: a history of sickness absence, and the patient's/employee's own prognosis of the duration of their sickness absence.

We invited all GP groups in Amsterdam to join this study; 29 GPs from 5 groups participated. They all used the same electronic information system (Arcos/EuroNed) containing basic information about their patients and their main problems, and a registration of daily contacts. Over the period July 1st 2000 until July 1st 2001 all registered patients in the age group 15–65 years were identified. All consultations with these patients were included in a sample if they contained the words: work, employment, business, company, factory, union, sickness absence legislation, sickness absence, absent, boss, manager, supervisor, human resource management department, personnel (for the Dutch terms see note 1) and the International Classification of Primary Care (ICPC) codes Z05: ‘problems with the working situation’ and Z08: ‘problems with social insurance’. In this way we included consultations with any work-related information. We read them carefully and removed the false-positives to get a final sample with a specificity of 100%, (e.g. in Dutch the abbreviation ‘ZW’ means sickness absence law as well as pregnancy). Next the consultations in this sample were analyzed independently by two researchers for the presence of information about current and former sickness absence and patients’ prognosis. Whenever interpretations differed, the two discussed until consensus was reached.

To control for false-negatives we analyzed the not-selected consultations with three GPs in the same patient group and period, and determined the consultations that should have been selected in. To compare the work-related consultations we constructed a study population as denominator, departing from all registered patients, age 15–65 years. We corrected this number with both 65% for having paid employment [8] and 70% for visiting the GP at least once a year [9], resulting in 14,515 patients/employees.

Outcome measures are the proportions of patients of the study population with consultations in a period of 1 year with any work-related information, with information about current sickness absence, prognosis of the sickness absence and the patients’ prognosis about duration.

Results

The control for false-negatives resulted in 7 patients out of 576 with consultations containing work-related information (for instance, ‘piano tuner’ was mentioned). So the sensitivity is 98.3%. The study population was 14,515 in which we found for 3,488 patients (24.0%) any work-related information, for 1,323 patients (9.1%) information about sickness absence, for 89 patients (0.6%) information about former sickness absence, and for 681 patients information about the patients’ prognosis of duration (4.6%) (Table I).

Table I.  Numbers and proportions of patients, estimated as having paid employment and visiting the GP at least once a year, with any work-related information in the registration system, and the presence of information about current sickness absence, former sickness absence, patients’ prognosis of duration.

	Number	%
The study population: registered patients, age 15–65 years, estimated as both having paid employment and visiting the GP at least once a year	14,515	100.0
Any work-related information present	3,488	24.0
Information about current sickness absence	1,323	9.1
Information about former sickness absence	89	0.6
Information about prognosis of duration of sickness absence	681	4.6

Discussion

Given the results that for 24% of the relevant group of patients GPs have any work-related recorded information, for 9% about sickness absence, for 0.6% about former sickness and for 4.6% about the patients’ prognosis of duration, we conclude that much information is missing for a systematic involvement of the GP. So a great deal of improvement is possible.

In another Dutch study analyzing consultations on videotapes, work was mentioned in 54% [10]. The difference between ‘mentioning’ and ‘recording’ may explain the difference in outcome. We found no other studies with which to compare our results. More systematic gathered information combined with a GP education program, can support a systematic follow up by the GP. It may lead to interventions that take these high risks into account, like referral to, and contact with the occupational physician (OP), in which the patient has to play an active role. In the context of a doctor-patient relationship characterized by mutual trust, there is a good chance that this policy will be supported by the patient and contribute in the long run to a reduction of LLSA.

With special thanks to Henk Brouwer and Jacob Mohrs for their help in the collection of the data and to Wendy van den Berg for her help in analyzing the data. This study was made possible by a grant from the social health insurance company Agis Zorgverzekeringen, The Netherlands.

Notes

1. Werk, arbeid, zaak, bedrijf, fabriek, vakbond, Ziektewet, ZW, ziekteverlof, verzuim, absent, arbo, baas, chef, PZ, personeel.