Introduction
Depressive disorders are common; the point prevalence is 5% (Angst, Reference Angst1992), and the lifetime risk for men and women is 27% and 45%, respectively (Rorsman et al., Reference Rorsman, Grasbeck, Hagnell, Lanke, Ohman, Ojesjo and Otterbeck1990). Depression is costly (Sobocki et al., Reference Sobocki, Jonsson, Angst and Rehnberg2006) and leads to great suffering (Coulehan et al., Reference Coulehan, Schulberg, Block, Madonia and Rodriguez1997). Most patients with depression can be effectively treated, and early detection should therefore be of great interest (Gelenberg and Hopkins, Reference Gelenberg and Hopkins2007).
The treatment prevalence among depressed individuals in the general population is low. In a survey of a defined geographic area (Jämtland County, Sweden) during 2001 and 2002, the point prevalence of depressed individuals was 4.5%, and 75% of them did not receive antidepressant medication (Henriksson et al., Reference Henriksson, Asplund, Boethius, Hallstrom and Isacsson2006). It has been estimated that 12% of patients in primary care suffer from major depression (Salokangas et al., Reference Salokangas, Poutanen, Kaartinen-Paschalis and Stengard1996; Bodlund et al., Reference Bodlund, Andersson and Mallon1999). This figure is actually probably higher as depressed patients often remain undiscovered (Coulehan et al., Reference Coulehan, Schulberg, Block, Madonia and Rodriguez1997; Lotfi et al., Reference Lotfi, Flyckt, Krakau, Martensson and Nilsson2010).
In a previous study of undetected depression in primary care, a two-step procedure was used with a screening [Montgomery Åsberg Depression Self Rating Scale (MADRS-S)] (Svanborg and Ekselius, Reference Svanborg and Ekselius2003) of all patients in the waiting room and a subsequent diagnostic interview of patients above the cut-off level (⩾12). A majority of the latter were verified to have depression using the diagnostic manual Primary Care Evaluation of Mental Disorders (Prime-MD) (Spitzer et al., Reference Spitzer, Williams, Kroenke, Hornyak and Mcmurray2000). It was concluded that 13% of all patients in the waiting room had an undetected depression (Lotfi et al., Reference Lotfi, Flyckt, Krakau, Martensson and Nilsson2010).
The reasons for not detecting depression have been addressed in several studies; there could be a lack of awareness of cues indicating depression or a lack of experience among healthcare professionals (Jameson and Blank, Reference Jameson and Blank2010). A lack of time and non-continuity in primary care may be other explanations. Updated knowledge seems to increase the rate of detection and treatment of depression, resulting in reduced suicide rates (Henriksson and Isacsson, Reference Henriksson and Isacsson2006; Szanto et al., Reference Szanto, Mulsant, Houck, Dew, Dombrovski, Pollock and Reynolds2007). Thus, in order to increase the rate of detection, the primary care physician's awareness of cues indicating suspected cases should increase. Signs such as anxiety, irritability, disturbed sleep, increased guilt and decreased concentration and libido have been suggested in studies as symptoms or cues that precede depression, but no study has yet focused on cues in undetected cases of depression (Fava et al., Reference Fava, Grandi, Canestrari and Molnar1990; Lotfi et al., Reference Lotfi, Flyckt, Krakau, Martensson and Nilsson2010).
The aim of the present study was to find ‘cues’ for an early identification of patients with undetected depression in primary care.
Materials and methods
The study is based on a previous study on the occurrence and severity of undetected depression in primary care (Lotfi et al., Reference Lotfi, Flyckt, Krakau, Martensson and Nilsson2010). It was carried out in a primary care centre of a community in the north-eastern part of Stockholm with a fairly healthy, well-educated and prosperous socio-demographic structure. The catchment area consisted of 9600 inhabitants with an even gender distribution and a predominant age range of 25–65 years (50%). As, in the previous study, it was found that 12% of the patients had undetected depression, the present study focused on possible cues noted in medical records during previous visits that differentiated patients with undetected depression from primary care patients without depression. In the previous study, after an informed consent, all consecutive patients aged ⩾18 years in the waiting room at a primary care centre were screened for depression during a 10-day period in December 2004 with the MADRS-S (Svanborg and Asberg, Reference Svanborg and Asberg2001). The previous and present studies were approved by the Medical Ethics Committee in Stockholm, Sweden (Dnr 04-761/2).
The case recruitment flow is depicted in Figure 1. Those who were already being treated for depression or had another severe mental disorder (eg, severe suicidality, personality disorder) were not eligible; they either remained on their prescribed treatment or were transferred to specialised psychiatric care. Patients with MADRS-S scores ⩾12 were identified as cases. Primary care patients in the waiting room the same day as the cases, and who scored below the cut-off level, were chosen as controls. Forty-five controls were selected from the 176 negative screens (MADRS-S < 12), being the first age- and gender-matched patient who filled in the MADRS-S after an identified index case. A criterion for both cases and controls was having had one or more previous contacts with the primary care centre during the past year. Cases who agreed to be interviewed (31 of 45 with a MADRS-S score ⩾ 12; 13% of all screened patients) were diagnosed using the Prime-MD (Spitzer et al., Reference Spitzer, Williams, Kroenke, Linzer, Degruy, Hahn, Brody and Johnson1994). The Prime-MD was performed within one week after the screening procedure. Seventeen (60.7%) and 11 (39.3%) of the cases were diagnosed with major (nine men and eight women) and minor (two men and nine women) depression, respectively (Lotfi et al., Reference Lotfi, Flyckt, Krakau, Martensson and Nilsson2010).
Case recruitment flow for patients with undetected depression. MADRS-S = Montgomery Åsberg Depression Self Rating Scale.
Primary care medical records for a period of one year preceding the screening and incident depression diagnoses were collected for the cases and the controls. One type of ‘cue’ was a sign indicating depression, defined as a note in the medical record made by the primary care physician that described at least one of the nine depressive criteria for major depressive episodes or associated features of depression, as defined in the DSM-IV-TR (American Psychiatric Association, 1994). Thus, the notes of the following symptoms were sought in medical records during the year preceding the index visit: Depressed mood; diminished interest or pleasure in all or almost all activities; body weight loss or gain or increase or decrease in appetite; insomnia or hypersomnia; psychomotor agitation or retardation; fatigue or loss of energy; feelings of worthlessness or inappropriate guilt; diminished concentration or indecisiveness; and recurrent thoughts of death or suicide. A number of physician-rated signs/symptoms were also noted such as tearfulness, irritability, brooding, obsessive rumination, anxiety phobias, excessive worrying over physical health and complaints of pain (eg, headaches or joint, abdominal, or other pains). If such notes were found for many visits, it was only recorded once for the same patient. An experienced psychiatrist (S.H.) performed the assessments of depressive signs and cues. The identity and diagnosis of the patient was not known to the psychiatrist.
Another ‘cue’ was decreased functioning, defined by a low score on the Global Assessment of Functioning (GAF) (American Psychiatric Association, 1994). These notes were sought in the medical records of both cases and controls. To determine function, the GAF was applied by a trained psychiatrist (S.H.) to register symptoms and function descriptions that were found in the medical records on any or all of the occasions that the patient had visited the primary care centre. The symptom and function values (eg, impaired work and social functioning) were assessed separately, and the lowest value was chosen according to the manual for GAF assessments. In case of many GAF values from different visits, the highest value during the year studied was chosen. The rationale for choosing the highest GAF value the previous year was to avoid false underestimations possibly related to the health care seeking situation.
Statistics
The data were analysed using the SPSS 12.0.1 software for Windows. Descriptive statistics were used. The Mantel–Haenszel odds ratio was calculated to measure the association between the presence or absence of cues and the occurrence of a MADRS-S score ⩾ 12. The statistical tests used for comparisons were the χ 2-test with Yates’ correction for continuity or Fisher's exact test. Student's t-test was used for continuous variables.
Results
Cases and controls
Of a total of 221 individuals (median age of 58 and a range of 18–91 years), 45 patients scored ⩾12 on the MADRS-S. Of these 45 patients (19 men and 26 women), the MADRS-S scores (mean ± SD) for men and women were 19.3 ± 6.8 and 18.9 ± 6.6, respectively (NS). Forty-five controls were selected from the 176 negative screens (MADRS-S < 12). Thus, the first age- and gender-matched patient was chosen who filled in the MADRS-S after the index case and who had visited the primary care centre at least once during the previous year. MADRS scores for cases and controls are shown in Figure 2.
MADRS-S scores for cases with undetected depression (n = 45; mean ± SD; 19.1 ± 6) and controls (n = 45; 3.5 ± 3.2) (P < 0.001). MADRS-S = Montgomery Åsberg Depression Self Rating Scale.
At least one depressive cue was found in 17 (38%) of the 45 cases compared with 8 (18%) of the 45 controls [odds ratio (OR) 2.81; 95% confidence interval (CI): 1.06–7.43]. The distribution of cues is shown in Table 1. In comparison, sleep disturbances were present in 11 (24%) of 45 cases and three (7%) of 45 controls (OR 4.53; 95% CI: 1.17–17.55). Other depressive symptoms were also numerically more common among cases, but the differences were not statistically significant. The GAF scores (mean ± SD) for the cases and the controls were 57 ± 6.5 and 63 ± 3.7, respectively (t = −2.6, P < 0.05).
Frequency of cues of depression in cases (n = 17; MADRS-S ⩾ 12) and controls (n = 8; MADRS-S < 12) in a consecutively recruited primary care sample (n = 221)
MADRS-S = Montgomery Åsberg Depression Self Rating Scale.
According to Prime-MD, there were 28 verified cases with major and minor depression, and therefore we performed an odds ratio calculation between those among the 28 patients who had cues of depression (n = 10) and the corresponding age- and gender-matched controls (n = 28) who had cues of depression (n = 3), and we still found a significant difference (OR 4.63; 95% CI: 1.11–19.26; P = 0.035).
Cues indicating depression had been present in one of the 11 cases with a Prime-MD diagnosis of minor depression and in nine of the 17 with major depression (9% and 52.9%, Fisher's exact test, one-sided, P < 0.05).
Discussion
The main finding was that cues of depression found in medical records were twice as common in primary care patients who screened positive for depressive symptoms compared with those who screened negative. The single most important cue was sleep disturbances (insomnia or hypersomnia), a finding in accordance with that of other studies (Riemann and Voderholzer, Reference Riemann and Voderholzer2003). Sleep disturbances have been suggested as a risk factor for depression in several studies (Dryman and Eaton, Reference Dryman and Eaton1991; Breslau et al., Reference Breslau, Roth, Rosenthal and Andreski1996). In a study of the course of remitted patients, self-reported sleep disturbance was found as a prodrome of relapse (Perlis et al., Reference Perlis, Giles, Buysse, Tu and Kupfer1997). Ohayon even proposed that insomnia may be ‘a ticking clock for depression’ (Ohayon, Reference Ohayon2007).
Patients with undetected depression had a lower functional status than those without. However, if a low level of functioning is present, it should be coupled with another cue to prevent over-diagnosing (Aragones et al., Reference Aragones, Pinol and Labad2006). The patient's level of functioning is an important health economic aspect of depression and should be followed up when evaluating the treatment result, as both discovery and improved function increase work productivity (Katzelnick et al., Reference Katzelnick, Kobak, Greist, Jefferson and Henk1997).
Other cues associated with depression were impaired work, loss of interest and fatigue. After the screening procedure, patients who were diagnosed with major depression had significantly more cues than patients with minor depression, suggesting that the presence of cues might qualify as a prognostic factor indicative of more severe pathology.
The primary care physicians did not act on cues they had noted in the medical records. They did not initiate any further investigation either by using screening or diagnostic interview methods, nor did they prescribe medication for any of the observed symptoms, for example sleeping medication for a sleep disturbance.
Cues of depression, whether in the form of a single symptom, sub-threshold depression or a manifest depressive episode, were frequently missed by the primary care physicians in the present study. Non-recognition was not limited to patients with a relatively mild, brief depression but was also evident in patients with a persistent depression, a finding in line with that of others (Van Os et al., Reference Van Os, Van Den Brink, Van Der Meer and Ormel2006). This indicates that there is a lack of knowledge or incentive to look more closely into symptoms that may be relevant for health and ability to function in the general population.
Apart from being cues of depression in a primary care setting, one or more depressive symptoms might be seen as cues, residual symptoms or prodromes of relapse along a continuum in depressive disorders (Judd et al., Reference Judd, Akiskal and Paulus1997). Fava and co-workers investigated such cues before the first onset of a major depressive disorder in psychiatric outpatients in comparison with controls (Fava et al., Reference Fava, Grandi, Canestrari and Molnar1990). Each of the patients reported at least one sub-syndromal cue before the onset of a depressive episode. Thus, one or more cues of depressive symptoms may be regarded as risk factors for a later depressive episode both in recurrent and first episode depression.
Limitations and strengths
The number of statistical analyses would normally require a correction for multiple analyses to avoid type 1 errors; however, we decided not do it partly because of the limited number of cases and controls but also because our findings are consistent with other studies assuring their validity (Dryman and Eaton, Reference Dryman and Eaton1991; Breslau et al., Reference Breslau, Roth, Rosenthal and Andreski1996; Riemann and Voderholzer, Reference Riemann and Voderholzer2003).
One of the strengths of this study is that the patients who were assessed for depressive symptoms were representative of the waiting room clientele, ranging in age from 18 to 91 years with a median age of 58, with 57.7% women. All the patients in the waiting room were invited to go through the screening procedure. Cases and controls were found in the same way. As the frequency of cues in the cases was compared with that in controls matched for sex and age, these potential confounders were controlled for. Furthermore, the assessment of the cues in the medical records was done by an independent investigator who was blinded to the screening results.
A weakness of the study was that the cues were identified in retrospect from notes made by the primary care physicians. The notes were regular notes in the medical record that were not recorded in a systematic manner. A prospective approach using a structured interview or screening method might have reduced the risk for bias. Nonetheless, by using regular ‘real-life’ medical records, it was revealed that the primary care physicians were aware of the symptoms but did not proceed with further actions. Another weakness is that the study was performed in one single site limiting the generalisability of the results. The findings are, however, consistent with those from other studies confirming their credibility (Fava et al., Reference Fava, Grandi, Canestrari and Molnar1990).
These findings, together with those in our previous study in the same patients, suggest that when certain cues (eg, sleep disturbances, decreased functioning and at least one sign of depression) are present, a subsequent two-stage opportunistic screening procedure may reveal both minor and major depressive episodes that otherwise would have been undetected (Lotfi et al., Reference Lotfi, Flyckt, Krakau, Martensson and Nilsson2010). In such cases, anti-depressive medication and cognitive behavioural therapy, combined or given separately, could be initiated to prevent prolonged functional disability and health problems (Kupfer et al., Reference Kupfer, Frank and Perel1989).
Conclusion
On the basis of the these findings, when cues of depression, especially sleep disturbances and decreased functioning, are present, implementation of a two-stage diagnostic procedure is suggested to improve the overall heath situation in a primary care clientele. Suggestions for future research include studying the health economic consequences in order to find both overall health-related incentives and financial incentives for such strategies. Furthermore, qualitative research, for example interview studies with GPs, may clarify the underlying reasons why depression remains undetected despite cues.
Acknowledgements
Dr Nils-Olov Sandberg and his staff at Waxholm PHC, Stockholm County Council, are gratefully acknowledged. The late Dr Svante Henriksson is acknowledged for his contribution to the study.
Funding
The study was financially supported by Lundbeck Pharmaceuticals. Svante Henriksson received a salary for his work. No other authors have been financed by this or any other source. The study was designed by the authors with no influence from Lundbeck.
Declaration of interest
The authors report no conflicts of interest. The authors are responsible for the design, content and writing of the article.
