An 18-year follow-up study of a lupus cohort in Shanghai


: Dr Shun-Le Chen, Department of Rheumatology, Renji Hospital, Shanghai Jiaotong University School of Medicine, 145 Shan Dong Rd(c), Shanghai, China 200001. Email:


Aim:  To investigate the long-term prognosis of systemic lupus erythematosus (SLE) and its risk factors.

Methods:  An 18-year (1980–1998) clustering follow-up study of 50 patients with SLE was performed.

Results:  The overall survival rate at 1, 5, 10, 15 and 18 years after the onset of illness was 98%, 98%, 84%, 76%, and 70%, respectively. The two major causes of death were infection and renal failure. Cox proportional hazard regression analysis showed that the presence of ≥ 7 of the American College of Rheumatology criteria for SLE at diagnosis and vasculitis were associated with worse survival. Thirteen (26%) patients were in remission with complete discontinuation of drugs, with the mean remission duration being 12 years (range: 2–17 years).

Conclusion:  It is possible for patients with SLE to get long-term remission through timely and effective treatment. Prevention of infection and renal failure and reducing dosage of corticosteroids by combined treatment with immunosuppressive drugs may enhance prognosis and survival in SLE.


Systemic lupus erythematosus (SLE) is an autoimmune disease of unknown aetiology with a wide variety of clinical and immunological features. Along with continuing development of diagnosis and treatment of SLE, the prognosis has improved significantly over the past decades, and the long-term curative effect has received more emphasis. In 1980, we showed clinical and immunological features of 50 SLE patients; after this, 5 years and 10 years follow-up studies on the same SLE patient group were performed in 1985 and in 1990, respectively.1–3 Herein we report the 18-year follow-up (1980–1998) of this same cohort of 50 patients, focusing on survival rate, quality of life, causes of death and risk factors affecting long-term prognosis.



Fifty SLE patients were all diagnosed at Shanghai Renji Hospital between August 1979 and September 1980, fulfilling 1982 revised American Rheumatology Association criteria for classification of SLE.4

Data collecting

All the medical records of the patients were documented. Laboratory data, including complete blood analysis, routine urine, 24 h urine protein, renal function, erythrocyte sedimentation rate (ESR), antidsDNA antibodies, antinuclear antibodies (ANA), precipitating antibodies to extractable nuclear antigens (ENA), were tested.

Estimation for disease activity and clinical stages

  • 1Disease activity of SLE was systematically evaluated using the SLE disease activity index (SLEDAI).5 Although it was published in 1992, all of its clinical and laboratory variables have been included in the clinic data protocol and have been collected prospectively from the inception of the cohort.
  • 2Classification of clinical stages for lupus disease was set up in 1980 by the Department of Rheumatology, Renji Hospital, based on the following description:2Active stage: SLE active signs were present including fever, malar rash, polyarthritis, serositis, major organ involvement elevated ESR or low concentration of serum complement (C3 or C4).Controlled stage: clinical symptoms disappeared and ESR is normal under the drug therapy.Stable stage: no symptoms and ESR is normal on low dosage of corticosteroid (prednisone < 15 mg/day), within half a year.Remission stage: discontinued corticosteroid or other immunosuppressive drugs for more than one year, and no relapse.

Statistical analysis

Results are expressed as mean ± standard deviation (SD), and risk ratio (RR) with 95% confidence interval (CI). Statistical significance was evaluated with Chi-square test or anova where appropriate. The outcome was mortality. Survival rates were calculated according to the Kaplan-Meier life table analysis method. The Cox proportional-hazard model was used in the multivariate analysis. A P-value of < 0.05 was considered the level of significance.


Demographic and clinical variables

Our follow-up study of 50 patients was finished in September 1998. In this group, 48 patients were female and two were male. Their mean age at onset was 24.5 ± 8.9 years (11–50), in which 88% (44/50) had their onset from 15–40 years old, 6% had their onset before 14 years and another 6% had their onset after age 40. At time of diagnosis, 30 (60%) patients had renal involvement, five patients had renal insufficiency (serum creatinine ≥ 120 µmol/L), 10 (20%) had cardiac involvement and four (8%) had central nervous system lupus.

Survival rate

By September 1998, 16 (32%) female patients died. The overall survival rates at 1, 5, 10, 15 and 18 years were 98%, 86%, 76%, 68% and 68% after diagnosis and 98%, 98%, 84%, 76% and 70% after onset, respectively.

Causes of death

The mean age of death in the 16 patients was 35.6 years ± 13.1 (20–68) and the duration of illness was 11.4 ± 5.2 years (1–24). In the first 5 years follow-up study, the main causes of death were renal failure 33% (3/9), infection 22.2% (2/9), gastrointestinal vasculitis 11% (1/9) and cerebrovascular accident 11% (1/9). At 18 years follow-up, the rate of death from infection increased to 31% (5/16), which including 80% (4/5) respiratory infection. The other causes were 31% (5/16) from renal failure, 12.5% (2/16) from cardiac failure, 12.5% (2/16) from cerebrovascular accident, 6.25% (1/16) from gastrointestinal vasculitis and 6.25% (1/16) from unknown causes.

Univariate analysis of survival in 50 patients

Univariate analysis demonstrated that fulfilment of ≥ 7 items of the ARA criteria (RR = 3.57), proteinuria (RR = 1.859), elevated creatinine serum level (≥ 120 µmol/L) (RR = 3.182), cardiac disorder (RR = 3.188), and vasculitis (RR = 2.333) at diagnosis were significantly associated with mortality (P < 0.05). However, thrombocytopenia, low concentration of C3 or C4, CNS involvement, antidsDNA antibody and Raynaud's phenomenon, had no significant association with poor outcome. Further multivariate analysis with Cox proportional-hazard model confirmed the significant negative influence of having ≥ 7 ARA criteria at diagnosis and vasculitis on SLE survival.

Current clinical status of patients

In this follow up study, 34 patients (68%) were still alive by the 18th year. Their mean age was 48.8 ± 8.1 years (34–66), with 23.7 ± 4.3 years (18–31) duration of illness from onset. Disease activity was evaluated in all patients who had survived, utilizing SLEDAI scores and these were classified according to clinical stage.

There were 26% of patients (13/50) in clinical remission with drug withdrawal for 12.3 ± 5.1 years (2–19), 10% (5/50) in the stable stage who take prednisone 2.5–10 mg/day. On the other hand, there are three (6%) patients in the controlled stage, two of whom continue to take prednisone 20 mg/day without a rheumatologist's prescription. Twelve (24%) patients are in the active phase, 10 of whom are not adherent to therapy, with eight of these having discontinued their medications. One case was not included because of end-stage renal disease without other active symptoms of SLE for 5 years, who was receiving dialysis and discontinued the usage of corticosteroids.

Avascular bone necrosis

Eight patients (16%) developed avascular bone necrosis after 6.8 ± 4.1 years (4 months to 14 years) of corticosteroid treatment. Seven of these occurred while being maintained on moderate dosages of corticosteroids (20–30 mg/day); only one patient received intravenous pulse methylprednisolone therapy.

SLE and pregnancy

Among the 48 female SLE patients in this cohort, 18 had undergone pregnancy after SLE diagnosis. The disease was aggravated during pregnancy in two (11%) patients but became controlled after induced abortion. The other 16 (89%) gave birth to healthy children, with none of these developing SLE symptoms to-date. Of the 21 mothers, two had died of renal failure 3 and 4 years after delivery.


Due to unknown aetiology, it is considered that SLE is a non-curable disease. Along with continuous development of diagnosis and treatment of lupus, the prognosis has improved significantly over the past decades, and long-term curative effect has received much emphasis. The 5-year survival rate has increased from < 25% in 1955 to > 90% in recent decades.6–8 Recently a 15-year survival rate has been reported among 63–80% of patients in Canada, US, Spain and Denmark.9–12 In 1995, Shakra et al. first reported from Canada that the 20-year survival rate was 68%.10 In this 18-year (1980–1998) clustering follow-up study of 50 Chinese lupus patients, we reported the overall survival rate at 1, 5, 10, 15 and 18 years after the onset of illness to be 98%, 98%, 84%, 76% and 70%, respectively, which is comparable to the data published around the world.

In 1985 Urowitz et al. reported a bimodal pattern in lupus mortality, which emphasized late mortality in SLE caused by atherosclerotic disease while early in the infection.13 Results from this study followed up to 18 years showed that major causes of death were infection (31%) and renal failure (31%). By reviewing reports over the last 10 years via PubMed, it is obvious that infection has become the first cause of death for lupus, which is more common than renal failure and cerebrovascular accident. Fatal infections included a high frequency of lung infection and septicemia.10–12,14–16 It should be emphasized that prevention of infection is still one of the key points for further improving the survival rate among lupus patients.

The major side-effect of corticosteroids is the increasing risk of infection in SLE patients and it is related to both dosage and duration. Another important side-effect is avascular bone necrosis which may seriously impair the patient's life-quality. In this 18-year follow-up study, eight patients (16%) had suffered avascular osteonecrosis from 6.8 ± 4.1 years (4 months to 14 years) after treatment with corticosteroids, and only a few patients received intravenous methylprednisolonum. This indicates that there is great variation among individuals in terms of the relationship between the incidence of avascular bone necrosis and corticosteroid dose and disease course. Besides infection and avascular osteonecrosis, corticosteroids have a number of other side-effects, such as GI bleeding, hyperglycemia and atherosclerosis. Austin et al. showed that incidence of renal failure in patients treated with corticosteroids for more than 10 years was as high as 90%,17 which may be associated with hypertension, hyperlipidaemia and renal fibrosis induced by corticosteroids. Thus the risk/benefit ratio of therapy should always be considered and individualized, and a minimized dose of corticosteroids need to be emphasized, which may contribute to long-term prognosis and quality of life for lupus patients. In survival analysis, multivariate analysis with Cox proportional-hazard model demonstrated that the fulfilment of ARA criteria ≥ 7 at diagnosis, and vasculitis in SLE, had significant influence upon poor prognosis as isolated factors, while other factors had additional effects only. It has been reported that thrombocytopenia was associated with poor prognosis in Caucasians, Blacks in America and Chilean lupus patients.8–10,18 However, we did not find such a tendency in our follow-up cohort, which may be associated with differences in race or limited patient numbers.

The incidence of SLE is more frequent in women during childbearing years. It was reported that during pregnancy the relapse rate in SLE was over 50% and the mortality rate was also higher.19 Petri thought there was heredity sensitivity in SLE as he carried out a follow-up study of 100 infants whose mothers were SLE patients, and found that two infants presented with SLE. Infants tended to have cardiac atrioventricular block when their mothers had anti-SSA and anti-SSB antibodies.20 In our 18-year follow-up study, 18 patients married and were pregnant after onset of SLE; 89% of these had normal childbirth and 63% did not have a relapse. No child was found to have SLE to-date. It was believed that women with lupus should not get pregnant because of the risks to both mother and unborn child, but now it has been commonly accepted that SLE patients may have normal childbirth with strict monitoring by experienced obstetricians and rheumatologists. The practice in our department suggested that pregnancy could be considered if the disease is inactive for at least 1 year, with oral prednisone doses < 10 mg/day.21

Among those who had survived by the 18th year of follow-up, 13 (26%) patients were in remission with complete discontinuation of drugs for an average 12 years (range: 2–17). These 13 patients need to be followed up for even longer periods.