Management of patients with persistent medically unexplained symptoms: a descriptive study

We analysed the longitudinal electronic medical record data of persons participating in an ongoing randomised controlled trial (RCT) called the CIPRUS study. The CIPRUS study aims to establish the effectiveness of treatment of undifferentiated somatoform disorder by a mental health nurse practitioner (MHNP) within general practice, versus usual care. The design of the CIPRUS study has been described elsewhere in more detail [19]. Potential participants were identified by running a search of the electronic medical records for patients who had consulted their GP at least twice in the previous 3 months with one or more complaints from the Robbins list [20]. The Robbins list consists of 23 physical symptoms that are associated with functional somatic syndromes. GPs then checked the selected patients to verify that these patients indeed had MUS according to them, and excluded patients who fulfilled one or more exclusion criteria. Exclusion criteria were: presence of a medical or psychological disorder explaining the symptoms, presence of a severe psychiatric disorder, currently receiving psychological treatment for MUS, having poor language skills or handicap that prevented patients from understanding the intervention. After patients were verified as having MUS by their GP, they were interviewed using a structured clinical interview (SCID-I) in order to determine whether they fulfilled the DSM-IV criteria for undifferentiated somatoform disorder (USD) [21]. Those fulfilling the criteria for USD were included in the current study. All participating patients gave written informed consent to extract data from their electronic medical records. In the current study, we used data from patients participating in the usual care group of the CIPRUS study. We used data from the group of MUS patients receiving usual care because we wanted to know which care patients received. Because we had already identified these patients as having MUS we used the data from this group. Obviously we could not use data from the intervention group, as the intervention consisted of a number of scheduled meetings with the MHNP within the general practice, which would also be recorded in the electronic medical records. This data would, therefore, not only reflect usual care, but also care provided due to being part of the intervention group of our trial.

Data were manually extracted by 3 researchers from electronic medical records of all participating patients between 21 November 2016 until 31 August 2017 in 30 participating general practices. Data were extracted for all MUS consultations in the 5-year time period for each patient prior to the search date. Data were collected from fields for prescription, medical tests and referrals, and from the GPs’ free text notes. Extracted data consisted of the date of the consultation, International Classification of Primary Care (ICPC) code corresponding to the consultation, and the management strategy of the GP, i.e. GPs’ own notes in the electronic medical records on what was carried out during the consultation and what they were planning or had arranged to do as a next step. For persons who were younger than 18 years of age during the 5-year time period, data were only collected from age 18 onward. MUS consultations were defined as consultations in which the GPs used ICPC codes that corresponded with the symptoms from the Robbins list (Table 1) [20]. Because there are no suitable corresponding ICPC codes for the symptoms ‘restlessness’ and ‘thoughts slower’, these two items from the Robbins list were omitted. Data from consultations that were not coded with ICPC codes corresponding to the Robbins list, but where the GP had noted ‘MUS’ or ‘somatisation’ were also collected.

After collection, the extracted data on management were categorised by one researcher (KS) according to the options for diagnosis and treatment in the current Dutch GP guideline [15]. The categories from the guideline used for classifying diagnostic strategies were exploration of all complaint dimensions, physical examination and additional diagnostic testing within and outside general practice (diagnostic referral). The categories used for classifying treatment strategies were shared problem definition, education and explanation, advice, treatment with medication, setting up a time contingent plan, scheduling follow-up appointments, referral to other primary care providers, and referral to secondary care [15]. Within primary care a patient can be referred to other care providers such as a (psychosomatic) physiotherapist or exercise therapist, mental health nurse practitioner, primary care social psychiatric nurse or primary care psychologist (e.g. trained in cognitive behavioural therapy) [15].

Data were analysed using SPSS version 22 for Windows. We used descriptive statistics to describe the study population and the management strategies. In order to determine whether there were any trends of providing various management strategies over time, we used cross-tabs and the chi-square test for trend.

Of the 77 patients, 80.5% were female. The mean age was 50 (SD: 17.1, range: 19–89). Over the 5 year period, the mean number of MUS consultations was 13 (SD: 17, range: 1–130), resulting in an average of 2.6 consultations a year. The symptoms patients presented with are provided in Table 1. The most frequently recorded symptoms were back pain (17,3%), weakness or fatigue (12,8%), extremity pain (11%) and sleep disturbance (10,2%). No consultations had codes for loss of appetite and gas or bloating. Seventy-one consultations (6,9%) were coded with codes other than those that appear on the Robbins list, however the GP had referred to MUS in these consultations. The codes used in this category were hysteria/hypochondria (ICPC code P75), neurasthenia/stress (P78), spastic colon/irritable bowel syndrome (D93), and ‘no disease’ (A97).

GPs varied in the way they recorded what was done during the consultations. This varied per GP as well as per patient and per consultation. Table 2 provides examples of data extracted from electronic medical records of three patients. There are 2 examples of brief records (≤10 words), 2 examples of medium-length records (11–30 words) and 2 examples of long records (≥31 words).The table also illustrates how these were categorised according to the GP guideline categories. An overview of the strategies the GPs used in the 1035 consultations is provided in Table 3. The most common diagnostic strategies were physical examination (24.5% of consultations, range among GPs 7.0–66.7%) and additional investigations within the GP practice (14.6%, range 0–50%). Of the additional investigations, laboratory tests such as various blood, urine and feces tests were done most frequently (11.1%, range 0–37.5%). Symptom exploration was recorded in 3.5% of the consultations (range 0–20.0%) and found among 40% of the GPs. Having administered the recommended symptom checklist enquiring about distress, depression, anxiety and somatisation symptoms, the 4-Dimensional Symptom Questionnaire (4DSQ) [22], was only recorded once (0.1%).

The most common treatment strategies were treatment with medication (24.6%, range 0–62.5%), followed by discussing progress (16.2%, range 0–41.5%), scheduling follow-up appointments (11.8%, range 0–33.3%), vitamin pills/injections (11.7%, range 0–36.8%, recorded by less than a quarter of the GPs, mainly in the same patients), providing education and explanation (11.2%, range 0–35.7%) and giving advice (10.8%, range 0–42.3%). Wait and see strategies were also recorded frequently (9.4%, range 0–40.0%). Medication requiring a prescription was prescribed most (at least 19.4% of all treatment strategies, range 0–40.0%). NSAIDs were prescribed most frequently (3.6% of all treatment strategies, range 0–13.5%), followed by psychopharmacological medication (3.4% of all treatment strategies, range 0–20.0%, recorded by almost half of the GPs) and opioids (3.0% of all treatment strategies, range 0–11.5%, also recorded by almost half of the GPs).

Referrals to a psychologist (0.5%, range 0–7.7%, recorded by 17% of the GPs) or a psychiatrist (0.1%), formulation of a shared problem definition (0.4%, range 0–7.7%, recorded by 10% of the GPs) and setting up a time contingent plan (0.1%) were management strategies that were used the least often. When referrals to secondary care were documented, it was often unclear whether the referral was for diagnostic or treatment purposes. Therefore, a category ‘referral to secondary care (unclear for diagnostics or treatment)’ was added. Finally, of the 44 management strategies categorised as ‘other’, GPs coded 28 consultations (2.7%) with ‘talk’, ‘listening ear’ and ‘encouragement’.

We conducted chi-square tests for trend for the largest categories of management strategies (n of consultations ≥75). Over the 5 year time period, there appeared to be significant trends in the course of ‘giving advice’ χ² (1) = 5.73, p = 0.017, ‘medication adjustment’ χ² (1) = 11.67, p = 0.001, ‘discussing progress’ χ² (1) = 11.31, p = 0.001, ‘scheduling follow-up appointments’ χ² (1) = 10.75, p = 0.001 and ‘contact if necessary’ χ² (1) = 4.11, p = 0.043. In all the above categories, the proportion of consultations in which the management strategies concerned were provided, increased over time. For all of the above management strategies except ‘contact if necessary’ there was a small decrease in percentage of the management strategy used within consultations after the first year, after which the percentages increased again. For ‘contact if necessary’, the percentage of the consultations increased steadily across time. There were no significant trends over time for the other management strategies.

The most frequent management strategies recorded by Dutch GPs included diagnostic procedures such as physical examinations and additional investigations, and therapeutic procedures such as prescribing medication, discussing progress and providing education, explanation and advice. Other strategies that focus more on listening to the patient and involving patients in their own diagnostic and therapeutic process, and decision making, such as ‘exploration of all complaint dimensions’, ‘shared problem definition’, and ‘shared plan for symptom management’ did not seem to be adopted as frequently. These latter management strategies are especially important for MUS patients [23]. Patients were also rarely referred to a psychologist or psychiatrist.

As the symptoms lasted longer, GPs tended to adjust medication more frequently, discuss progress more often, schedule more follow-up appointments and encourage patients more to contact the practice if necessary.

When comparing these strategies to the recommendations in the Dutch guideline, we can conclude that GPs partly used management strategies recommended by the guideline but several essential strategies were missing. This may possibly reflect either the GPs’ or the patients’ reluctance to seek mental health care for complaints that are perceived to be primarily physical. However, another reason could be that GPs in our sample were not sufficiently familiar with the guideline yet, since the guideline was published during the data extraction period. Even if GPs were familiar with the guideline, it may have taken some time to get used to the new approach, and they may not have started applying strategies, such as exploration of all complaint dimensions, with patients whom they had already seen often before.

Our findings could, however, also point to underreporting of these, more ‘conversation-like’ management strategies. Recording behaviour varied widely across GPs, so it is impossible to know whether the strategy was not provided or not recorded. Due to time constraints, GPs may only put the more objective management strategies such as results of physical examinations, additional investigations and medication prescriptions in the medical records.

Several studies investigated management of MUS in other countries. A Norwegian study found that the majority of Norwegian GPs offered supportive counselling (64%), followed by prescribing medication (24%) and additional tests or referrals (20%) [8]. In our sample, the rates for prescription of medication (24%) and additional testing or referral (18%) were similar.

An Italian study found that Italian GPs mostly provided reassurance and support, listened to the patient, prescribed medication, ordered further medical tests and provided information [12]. In our study, prescribing medication, doing further testing and providing information were also among the most commonly used strategies, however offering reassurance and support and listening to the patient were recorded less frequently. Although the GPs in our study coded 2,7% of their consultations as ‘having a talk’, ‘listening ear’ and ‘encouragement’, ‘listening to the patient’ was not one of the categories that we used in our classification of management strategies. Also, not all GPs may record their listening behaviour as such in the medical records.

In the dental field adherence to clinical practice guidelines has been found to be up to 72% on average [24]. GPs also do not fully adhere to clinical practice guidelines [25, 26]. GPs report that they are aware of the guidelines, but find it difficult to implement them with all individual patients, as they may feel that the guideline is not always fitting. GPs may therefore prefer to provide personalized care and let the patients have the final say in their treatment [25, 27]. This may also apply to the GPs in our study.

To the best of our knowledge this is the first study to investigate care for persons with MUS in such detail. A strength of this study is that we used real-world data directly from electronic medical records. We were therefore able to collect detailed information about every MUS consultation. Furthermore, we did not rely on self-report instruments such as surveys or interviews taken from GPs. This possibly led to having gathered more ‘objective’ data, free from various kinds of bias such as recall bias. Another strength is that the choice of categories for classifying management strategies was based on the current Dutch guideline, which provided clear classification options beforehand. At the same time, it must be noted that the guideline is a best practice statement, which is based on meta-analyses of high-quality randomized controlled trials where possible, but is not always the case. As a part of the recorded consultations took place before the guideline was published, a longer period of time is needed to draw firmer conclusions regarding adherence to the guideline.

Another limitation of this study is that by using electronic medical records our data were completely dependent on the registering behaviour of the GPs, which varied in amount of detail and coding. If the GP did not record certain management strategies or symptoms in a consultation, these data were missing. Our data, therefore, do not necessarily reflect what was actually done during the consultation, rather what was done and recorded. A comparative study with recordings of patients with and without USD, and comprehensive recording of all types of management strategies by GPs or videotaped consultations would be helpful in gaining thorough insight in their management and subsequent recording [28, 29].

A final limitation is that it was not possible to decide which consultation was the first consultation in the course of one or more MUS episodes. Because of this, all consultations were analysed as if they are independent. However, this is usually not the case. The policy of the GP can depend on the findings and results from previous consultations.