Electronic searches were performed using the online databases of PsycINFO, Embase and PubMed from inception to 5th May 2022. Search terms are detailed in Table 1. Reference lists of relevant publications were manually checked to identify potential studies related to DD and SZ.

This yielded 5,900 records. Subsequent to the removal of duplicates, studies were screened for eligibility by titles and abstracts, and then by full texts (Figure 1). The search strategy was performed by three independent authors (LC, PH and CH). Any disagreement among the authors was resolved through discussions.

Articles were included if they met the following criteria: (1) included comparative data in the clinical, cognitive or functioning outcome of DD and SZ; (2) consisted of patients with DD and SZ according to ICD or DSM criteria; and (3) published in an English, peer-reviewed journal.

Articles were excluded if they were: (1) case reports, systematic reviews, protocols, conference abstracts, commentary or meta-analyses; (2) not comparing between DD and SZ; or (3) included patients without a clear description of the diagnostic criteria for DD and SZ according to the ICD or DSM.

Titles and abstracts of retrieved publications were reviewed by three independent authors (LC, PH and CH) to determine relevance. Where titles and abstracts failed to provide sufficient indication of relevance, full articles were examined for eligibility with regards to the abovementioned inclusion and exclusion criteria. Studies included in the systematic review were then analyzed with reference to statements about clinical, cognitive or functional differences between DD and SZ.

For each included study, the following variables were recorded: authors and year of publication, title, objectives, study design, study setting, location of study, participants’ age, onset age, diagnosis and its change over time and the outcome measures used. Main findings on the clinical, cognitive or functional outcomes between DD and SZ were presented separately for non-age-matched and age-matched samples, if available.

For cross-sectional studies, the Joanna Briggs Institute (JBI) Critical Appraisal Checklist for analytical cross-sectional studies was used (12). One out of eight items were removed because of its irrelevance to the studies included (Table 2). The appraisal tool assessed the methodological quality of a study and addressed the possibility of bias in its design, conduct and analysis based on sample inclusion criteria, study setting, condition measurement, confounding factors, outcome measurement and statistical analysis. Each item was addressed with “Yes,” “No” or “Unclear.”

For longitudinal studies, the JBI Critical Appraisal Checklist for cohort studies was adopted (15). Four out of 11 items were removed since the questions were not relevant to the studies included (Table 2). The checklist assessed the methodological quality of a study and addressed the possibility of bias in its design, conduct and analysis based on confounding factors, outcome measurement, statistical analysis and follow up strategies. Each item was addressed with “Yes,” “No” or “Unclear.” For both types of studies, the risk of bias was ranked as high when “yes” scores were ≤ 49%, moderate when “yes” scores were between 50 and 69% and low when “yes” scores were above 70%.

Of the 5,900 articles initially retrieved, 1,530 were identified in PsycINFO, 1,363 in Embase and 3,007 in PubMed. Following title and abstract screening, 31 articles remained for full-text retrieval, one of which was excluded for lacking documentation of a full article. Of the remaining 30 articles, 22 were excluded: one was not published in English, nine were excluded due to study design, three did not evaluate clinical or functional variables in DD and SZ, eight did not compare between DD and SZ, and one included participants without a clear description of the diagnostic criteria for DD and SZ according to the ICD or DSM. In total, eight studies fulfilled our inclusion/exclusion criteria and were therefore included in the systematic review (Figure 1). Two of those studies shared the same sample pool (16, 17). Characteristics of the included studies were summarized and discussed in Table 3.

Of the eight included studies, six were cross-sectional studies and two were longitudinal with a follow-up period of 13 years (5) and 7 years (18). Four studies were compiled and published in Spain, one in Canada, one in the United States, one in Germany and one in Hong Kong. Four out of eight studies recruited patients from outpatient clinics, three recruited from inpatient facilities and one recruited from a university medical center as well as the community.

Three of the eight studies compared between DD and SZ (6, 17, 19). Two studies compared between DD, SZ and schizoaffective disorder (16, 17). One study compared between DD, paranoid SZ and non-paranoid SZ (20), one study compared patients with late-onset SZ, DD with hallucination and DD without hallucination (18), and one study compared between DD and paranoid SZ (5). DD and paranoid SZ patients in the last study were diagnosed according to ICD and DSM criteria, while the diagnosis in all other studies was made according to DSM criteria. This comprised of a total of 585 DD patients, 1,124 SZ patients and 63 schizoaffective disorder patients.

Four studies conducted age-matched comparisons. Evans et al. (14) selected only patients with illness onset after aged 40 to produce an age-matched cohort. The mean onset age was 60.4 years for DD patients and 54 years for SZ patients.

Hui et al. (6) identified an age-matched cohort by propensity score matching, including DD patients with a mean age of 39.4 and SZ patients with a mean age of 39.1 at first episodes.

Two other studies (19, 20) performed age-matched comparisons between DD and SZ patients by statistical age-adjustment. In one study, the mean onset age was 38.8 for DD patients, 30.5 for paranoid SZ patients and 23.9 for non-paranoid SZ patients (20). The age of onset was not documented by Muñoz-Negro et al. (19).

Three of the four aforementioned studies conducted non-age-matched comparisons as well.

The remaining four studies (5, 16–18) performed only non-age-matched comparisons between non-matched DD and SZ cohorts.

Two out of eight studies documented diagnostic change over time. Over the follow-up period of up to 8 years, none of the DD or SZ patients had a change in their primary clinical diagnosis (14). Meanwhile, another study recorded 21.2% of the DD patients shifting into SZ or schizoaffective disorder during a period of 10.8 years (5). The remaining 78.8% of DD patients had no syndrome shift.

The effect of age was apparent on positive and general psychopathology symptoms, but absent for negative symptoms, functioning and hospitalizations. Regarding positive symptoms and general psychopathology, DD patients were found to have more severe symptoms in age-matched cohorts but less severe symptoms in non-age-matched cohorts when compared to SZ patients. With existing research suggesting younger age to be associated with a poorer prognosis in SZ (10, 11), one may reasonably expect DD patients who are generally older (3, 4) to have less severe positive and general psychopathology symptoms than SZ patients in non-age-matched comparisons.

However, the effect of age on negative symptoms and hospitalization remains ambiguous. In accordance with other studies indicating DD patients to be characterized by less pronounced negative symptoms like flat affect and alogia (4), we found the DD displayed less severe negative symptoms irrespective of age-matching. This may be because studies that included only outpatients were biased towards clinically less severe samples, particularly in DD patients, leading to more notable differences between DD and SZ. Though unable to conclude DD and SZ as completely separate entities, our results reiterated dissimilarities between the two groups and suggested the possibility of a psychopathological gradient regarding negative symptoms among psychotic disorders.

Similarly, we found DD patients to have fewer hospitalizations regardless of age-matching. Existing studies that reported non-age-matched DD cohorts as having fewer hospitalizations may reflect the observation of a better prognosis for DD patients who tend to be older (5, 13). Our consistent findings across age-matched and non-age-matched studies, however, challenged this explanation considering the minimal effect of age on hospitalizations. The fact that most DD patients were hospitalized due to social reasons (5) may imply that they were less disturbed by clinical symptoms in the first place, accounting for fewer hospitalizations in general. It should also be noted that most of the existing studies did not explore reasons of hospital admissions. It would therefore be worthy to compare reasons of admissions such as relapse, suicide, or comorbid health conditions. Moreover, very few studies looked at voluntary and involuntary admissions. Further investigation on the types of hospital admission and its relationship with help-seeking behaviors or insight would be meaningful. Despite the absence of age effect on negative symptoms and hospitalizations, the inconsistent findings regarding positive and general psychopathology symptoms reveal how the effect of age on clinical characteristics remains pivotal.

Given that an older age of onset was closely linked with better prognosis in psychotic disorders (10, 11), it is reasonable to expect aged- and non-age-matched differences in functioning outcomes between DD and SZ.

In non-age-matched samples, our results consistently pointed towards better global, social and occupational functioning in DD patients. This observation was in line with our expectation that in comparison to their SZ counterparts, DD patients would be more likely to manifest better functioning due to their older age.

However, further investigation into studies that compared both aged-matched and non-aged-matched cohorts may provide more important clues as to the impact of age on these outcomes. For instance, Hui et al. (6) found better occupational outcome in DD in non-aged-matched analyses, but the same study did not observe such a difference when patients were matched by age, implying the substantial impact age has on functioning outcomes. Meanwhile, Peralta and Cuesta (20) found age-matched and non-age-matched DD patients to consistently demonstrate better functioning, potentially due to the age adjustment method adopted.

As for cognitive functioning, both age-matched and non-age-matched studies consistently found no difference between DD and SZ. However, it should be noted that only two studies looked at neurocognitive functioning, rendering insufficient data in concluding with certainty that DD and SZ do not differ from each other in terms of cognitive functioning.

While age influences gender, education and premorbid functioning between the two groups, it has little to no effect on marital status. In line with previous studies that reported DD patients with less deterioration of social, intimate and established relationship before onset (3, 4, 21), we found that more DD patients were married regardless of age-matching. Additionally, the mean age of onset for DD patients in our included studies was above 40. Since the illness occurred during middle-to-late adult life, it is more likely for DD patients to have been married by the time they fell ill. The consistent findings across age-matched and non-age-matched studies thus diminished the effect of age on marital status.

Nevertheless, we observed inconsistent results for other demographic variables. While DD patients were found to be less educated and were predominated by women in non-age-matched cohorts, no differences in gender, education and premorbid functioning were recorded in age-matched cohorts. In view of the discrete results, it is plausible to speculate an effect of age regarding the aforementioned variables.

While the consistency of findings was generally good across studies, it was difficult to conclude the relative impairment between DD and SZ in several of the outcomes. For instance, ratings of general psychopathology between age-matched cohorts tended to be higher for DD in two studies (6, 14), but the opposite was observed in another (19). This may be related to the large discrepancies between the studies reviewed, as not all had matched for age. Further, some of the non-age-matched studies recruited only inpatients (5, 18, 20), some combined samples from five independent studies (16, 19) and one recruited DD and SZ patients at slightly different periods (20). That the majority of the samples were not truly representative makes age-matching of paramount importance. Additionally, longitudinal studies are crucial in identifying the distinctions between DD and SZ in the long term, but the two included were both non-age-matched and recruited only inpatients. They only provided enough information to conclude DD and SZ inpatients to be distinct from each other when not matched for age.

Furthermore, methods of age-matching varied across studies. One study conducted propensity score matching (6) while another selected only patients with illness onset after age 40 (14), which may have led to fewer SZ cases in the sample. Some performed statistical age adjustments (19, 20) which may not have yielded accurate measures of actual differences. Given the above discrepancies, a truly representative sample accurately matched by age may be needed for comparison between DD and SZ.

The directness of evidence is also limited by the discrepancies in patient groups across studies. For instance, some studies only involved patients with DD and PSZ (4) while others also included SA groups in their comparisons (16, 17). As previously mentioned, some of the studies also only recruited inpatients while others only outpatients. Recruiting DD inpatients may create a bias towards admission due to functional reasons instead of sheer clinical symptoms (5), with a greater severity of symptoms across all inpatients. Therefore, outpatients should also be included to secure a more representative sample of DD, especially given the questionable accuracy of hospital admission data regarding the true occurrence of DD in the population (3).

Further detracting from directness is the discrepancies in outcomes measures between studies. While most of the studies assessed clinical characteristics using the Positive and Negative Syndrome Scale, one study (20) adopted the Comprehensive Assessment of Symptoms and History. Another study (18) did not indicate the clinical scales used in their assessment. Assessment materials for functioning also varied across studies, with some adopting Social and Occupational Functioning Assessment Scale, some using Global Assessment of Functioning and some Disability Assessment Scale.

Similarly, diagnostic classification systems varied across studies. While the majority of the studies made diagnosis according to DSM criteria, one study had patients diagnosed according to ICD and DSM criteria. Additionally, while most of the studies that adopted DSM criteria diagnosed according to DSM-IV, two opted for DSM-III which may now be considered outdated. Using uniform and updated diagnostic classification systems shall thus be useful to ensure diagnostic categorizations are met.

Multiple comparisons were not always controlled for in the included studies. Some studies (6, 17) took the problem of multiple testing into account, but some (17, 20) did not. This might inflate the possibility of Type I errors which may cause an overestimation of existing differences between DD and SZ.

With the primary focus being on English literature, not all available data pertaining to the topic may have been identified. Conclusions about the differential outcomes of DD and SZ may consequently be underestimated especially when applied to non-Western countries. Further selection bias may have been introduced by including three studies that were conducted by the same authors (16, 17, 19), in which one (19) partially derived their data from the other two independent studies (16, 17). Conclusions about the quality of the studies reviewed may also be limited as the relevant authors were not contacted for clarifications, despite one study neglecting to state the materials they used to measure clinical outcomes (18).

Neurobiological research into the cellular and molecular mechanisms underlying psychotic disorders may provide additional insight about the nosologies of DD and SZ. Previously, gray matter reductions in the superior temporal gyrus and cerebellum were indicated as neuroanatomical markers of psychosis (22). Future research may be guided by the Research Domain Criteria project to identify genes, cells, and other units of analysis associated with the superordinate functional constructs of psychotic disorders, such as negative and positive affect, cognition, and social processes (23). In this way, neurobiological advances may help to further refine diagnostic classification beyond observable characteristics, and better account for the outcomes of DD and SZ.

As of now, current evidence suggests that DD and SZ demonstrate similar levels of cognitive impairments regardless of age. Cognitive treatments that have recently been recommended for SZ (24) may thus also be applicable to improving the cognitive performance of patients with DD. Of particular relevance is cognitive remediation therapy, which offers benefits across different cognitive domains including memory, planning, problem solving and social cognition, independently of age (25).

There is currently insufficient evidence to conclude whether DD is completely distinct from SZ. Our systematic review has found extensive effect of age on positive and general psychopathological symptoms as well as functioning, but consistent differences between DD and SZ in terms of negative symptoms and hospitalizations regardless of age matching. From these we can only infer that DD and SZ exhibit dissimilarities regarding negative symptoms and hospitalizations at the time of data collection. Moreover, since too few studies explored cognitive functioning, there is insufficient empirical data to determine whether SZ and DD patients differ from each other in terms of cognition. Another reason for the evasive conclusion would be due to difficulty in ascertaining enough DD sample. Given DD accounting for less than 1% of hospital admissions (3), studies that recruited inpatients only might end up with a small DD sample size. Also, many studies did not control for multiple comparisons, therefore existing differences between DD and SZ might be overestimated.

Our systematic review supports that age is an important prognostic factor in SZ, future studies should thus bear in mind the confounding effect of age when comparing different psychotic disorders with different ages of onset. As mentioned above, whether statistical age adjustments are accurate measures of actual differences remains questionable and that the selection of patients with older onset age would bias towards fewer SZ cases. When DD and SZ patients of all age groups are recruited during a particular period for comparison, the problem of neglecting adolescent-onset cases would be minimized. Alternatively, future studies should at least include outpatients to ascertain a more representative DD sample, while making sure to collect longitudinal information which is crucial to determine whether DD is distinct from SZ in terms of course and outcome in the long term. Future studies should take into account other confounding variables such as cultural differences. For instance, there is lower cannabis use in Asia (<0.5%) than in the Western countries (>10%) (26), thereby influencing risk factors such as substance abuse before onset. Also, the fact that Chinese population who suffers from serious mental illness tend to demonstrate more self-blame (27), might give rise to reduced openness domestically and less prevalent professional help seeking behaviors.